# Changelog
Source: https://docs.codebanana.com/changelog/changelog
Stay current with CodeBanana product updates, release notes, new features, improvements, bug fixes, and recent platform changes
## **Team Plan updates**
This release introduces several enhancements to **Team Plan**, making billing, quota management, and administration more flexible for organizations.
### **Shared token pool**
Team Plan now supports a **shared token pool**.
For existing Team Plan organizations, any remaining included tokens will be automatically migrated into the shared token pool during the upgrade. This allows all eligible members to consume tokens from a centralized balance instead of individual allocations.
### **Batch member management**
Team Plan now supports **batch configuration** for all member settings.
Admins can configure multiple members at once, including token limits, available models, VM configurations, and other Team Plan settings, significantly reducing management overhead for larger organizations.
### **Model management improvements**
Model management has been expanded with reusable configurations.
Admins can now:
* Save model lists as reusable templates
* Apply model templates to members in bulk
* Configure the default model for each template, including whether **Auto** mode should be used as the default
These updates make model governance more consistent across organizations.
### **VM downgrade support**
Virtual machines can now be **downgraded** directly within Team Plan.
Organizations can adjust VM specifications more flexibly as team requirements change.
### **Team Plan renewal improvements**
Team Plan renewals now support payment directly from the **Budget Balance** (shared token pool).
Previously, Budget Balance could only be used for adding seats and upgrading virtual machines. It can now also be used when renewing Team Plan subscriptions, providing a more unified billing experience.
## **Personal Agent, Project Management, and Message Experience Updates**
This release includes major improvements to **Personal Agent**, project customization, model availability, member management, message forwarding, and multiple stability improvements.
## **Project avatar support**
Projects now support custom avatars.
Users can configure project avatars from:
**Chat area → Top-right “…” → Settings**
This provides more flexibility for identifying and managing projects across different collaboration scenarios.
## **Personal Agent improvements**
**Personal Agent** has been significantly enhanced with broader project and usage awareness.
Personal Agent can now query and summarize:
* Project data, member information, and newly joined projects
* Personal usage and consumption data
* Team Agent and Private Ask messages (previously only Discussion messages were supported)
* Daily, weekly, and monthly summaries
* Project activity data
* Usage breakdown by model
These updates make Personal Agent more capable as a personal workspace assistant for understanding project progress, activity, and usage patterns.
## **New model support**
CodeBanana now supports additional models:
* **GPT-5.6-sol**
* **GPT-5.6-terra**
* **GPT-5.6-luna**
* **Claude Sonnet 5**
These new models expand available choices for different development scenarios.
## **Members dialog improvements**
The Members dialog has been updated with improved management capabilities.
New updates include:
* Improved admin configuration for group administrators
* Agent access permission settings
* Robot management through **Add Robot**, including webhook URL display and copy support
* Improved **Add Agent** dialog with an **Added Agents** filter category
Additional visual improvements will continue in future iterations.
## **Message forwarding improvements**
Message forwarding now supports **i18n** and quoted messages.
Forwarded messages now preserve original reference relationships and support multilingual content handling, making shared conversations easier to understand across different languages.
## **Agent and Skill improvements**
Fixed an issue where HTML-related tasks did not prioritize loading the **single-file-html Skill**.
HTML generation tasks now consistently trigger the HTML Skill by default, improving HTML generation quality.
Fixed an issue where the Read File tool could not correctly process **TicNote** share links.
TicNote `shareDetail` links are now automatically converted to the correct `shareDetailReal` format for reading.
## **Bug fixes and stability improvements**
This release includes multiple fixes across messaging, sharing, surveys, avatars, and references.
* Fixed an issue where forwarded messages could contain an incorrect `message_id` and display the previous message instead
* Fixed an issue where HTML detail content could return empty during sharing scenarios
* Fixed an issue where survey return page links could contain unnecessary language path prefixes
* Fixed group avatar validation logic to support empty `avatarUrl` values in text-based avatar mode
* Fixed an issue where quoted messages from previous pages were not displayed correctly
These improvements enhance message consistency, sharing reliability, and overall product stability.
## **Internationalization, Model Updates, and Mention Experience Improvements**
This release includes updates across internationalization, default model settings, message forwarding, @Mention experience, navigation interactions, file tree display, and GitLab project creation.
## **Internationalization updates**
CodeBanana now supports language switching across key user-facing pages.
Users can now switch between **English** and **Chinese** from the login page, homepage header, and Personal Center. Main-level pages now support Chinese display for a more localized experience.
## **Default model update**
The default model for new projects has been updated to **GLM-5.2**.
New projects will now use GLM-5.2 as the default model, while users can still select other available models based on their needs.
## **Message forwarding improvements**
The forwarding experience has been improved across message display and interaction.
Updates include:
* Improved forwarding titles and message prompts
* Added source group information to forwarded messages
* Fixed an issue where chat history and images in forwarded messages could be incorrectly clickable
These updates make forwarded conversations easier to understand and navigate.
## **@Mention improvements**
The **@Mention** experience has been significantly improved across file selection, input behavior, and stability.
Files in the @Mention panel are now displayed as a file tree by default, making project structure easier to browse. Skill filtering has also been improved, and highlight colors are now aligned with the App experience.
Input behavior has been optimized with fixes for multiple scenarios, including:
* Incorrect line breaks when using Command + Enter
* @Mention not triggering correctly
* Incorrect insertion position after mentioning
* Project names containing special characters not being recognized
* Unexpected @ triggering behavior at the beginning, middle, or end of text
Additional stability improvements include fixes for React #300 errors in the Mention system, popup positioning issues, cursor calculation problems involving BR tags, ineffective line breaks after @Mention, and occasional crashes.
The @Mention project list now also refreshes in real time.
## **Interaction improvements**
Navigation and switching behavior have been optimized.
Debounce handling has been added to **Sidebar**, **ChatList**, and **ProjectList** to reduce unnecessary requests caused by frequent switching between chats and projects.
## **File tree improvements**
The file tree interface has been visually optimized.
These updates improve readability and make project file navigation more consistent.
## **GitLab project creation improvements**
Fixed an issue where saved passwords could become unavailable after logging out of GitLab during project creation.
GitLab project import and creation flows are now more stable.
## **Team Plan, Organization Management, and Agent Updates**
This release includes updates across Team Plan member management, model controls, organization contacts, onboarding experience, Agent capabilities, IM integrations, and multiple stability improvements.
## **Team Plan member management and model settings**
Team Plan now supports more flexible member-level configuration.
Organizations can now configure different available models and token limits based on organizational structure from:
**Usage → Members & Quotas → Batch Settings**
Admins can also manage model configurations from:
**Usage → Members & Quotas → Manage Models**
Supported capabilities include:
* Save model lists as templates
* Customize model ordering
* Set default models for members
This release also fixes an issue where Team Plan members with a token limit set to **0** could not have their limit updated afterward.
## **Organization contacts and management updates**
The Contacts experience has been improved to support more complex organizational structures, including cases where one person manages multiple departments or members within the same department have different direct managers.
New capabilities include:
* Set department heads from **Contacts → Department list → Edit**
* Set direct managers from **Contacts → Member list → Edit**
* Display **Direct Manager** information in member profile cards
These updates provide better visibility into reporting relationships and organizational ownership.
## **System notification updates**
Global system message types have been expanded.
Organization join notifications and friend invitation notifications are now displayed globally and are no longer restricted by organization filtering.
## **Survey page localization**
Survey-related pages now support multilingual display.
This improves accessibility for users across different language environments.
## **Friend invitation notifications**
Friend invitation flows now include two-way notifications.
When a friend invitation is accepted or rejected, the original sender will receive a corresponding system notification.
## **Model selection persistence**
CodeBanana now remembers the model selected by the user.
When returning to the product, the previously selected model will be automatically restored.
## **Preset project configuration updates**
Preset projects now support parameter-level configuration.
Different preset project types can now define whether users can invite members, add agents, and other project-specific behaviors.
Preset projects also now support additional protections:
* Preset projects cannot be deleted or dissolved
* Preset project avatars have been improved, including **TN Subscription**, **Personal Agent**, and **Approval Workflow** projects
## **Personal Agent updates**
The **Personal Agent** experience has been improved.
Updates include:
* Improved avatar display
* Expanded data query scope
Personal Agent can now query messages sent across **Team Agent**, **Private Ask**, and **Discussion** contexts.
Historical Personal Agent skills are still being updated progressively.
## **Agent updates**
Agent workflows now include additional safeguards and improvements.
Cross-project Discussion delegation is now restricted to ensure correct project-level chat boundaries and prevent unintended context sharing.
Scheduler cost tracking now includes task creator information, improving usage attribution and cost visibility.
## **IM integration updates**
Several improvements have been made to IM channel integrations.
Updates include:
* Fixed and optimized channel icons
* IM pages now retain selected engine and model settings
* Fixed message reply limitations for QQ, Telegram, and other supported channels
## **Discussion reply improvements**
Fixed an issue where Discussion could not correctly respond when users referenced quoted messages.
Discussion replies now handle referenced content more reliably.
## **Friend invitation notification fixes**
Fixed an issue where friend invitation accept/reject notifications could contain incorrect organization IDs.
Notifications are now correctly routed to the intended users.
## **CBBot project permission fixes**
Fixed an issue where members joining CBBot-created projects through invite links or direct invitations could incorrectly receive admin permissions.
New members in CBBot-created projects now join with standard member permissions by default.
## **Private Ask queue display fixes**
Fixed an issue where **Private Ask** could display other users’ queue information while waiting.
Users now only see their own queue status during request processing.
## **Image Generation, Approval Workflow, and Billing Updates**
This release includes updates across agent capabilities, image and video generation, Cron Jobs, approval workflows, billing management, onboarding, contacts, context management, and model availability.
## **Image and video generation**
CodeBanana now supports **image and video generation** capabilities within Agent workflows.
Agents can now generate visual content directly during conversations. Related usage tracking and billing logic have also been added to the **Usage** page, providing clearer visibility into generation usage.
## **Cron Job updates**
Cron Job management has been improved.
Users can now specify the task creator and select the model used for scheduled tasks, providing more control over automated workflows.
## **Approval Workflow**
CodeBanana now includes the **Approval Workflow** project.
Organization **Owners** and **Admins** can configure approval workflows, while organization members can view and submit approval requests.
This enables teams to manage internal approval processes directly within the workspace.
## **Billing mode settings**
A new **Billing mode** setting is now available under:
**Personal Center → Billing mode**
Organizations can now choose whether usage costs are calculated based on the **presenter** or **owner**.
This setting is available to **Owners** and **Admins**, providing more flexibility for organization-level cost management.
## **Billing improvements**
Preset project bootstrap operations are now excluded from billing.
This means initial project setup processes will no longer consume user quotas or generate usage charges.
## **Onboarding improvements**
Organization onboarding now includes duplicate organization name validation.
This helps prevent naming conflicts when creating or joining organizations.
## **Contacts updates**
The **Contacts** page has been improved with better organization and contact management.
Contacts can now be displayed separately by internal and external groups. The page layout has also been optimized, and contacts can now be sorted by employee ID and department.
## **Context Clear improvements**
The **Context Clear** experience has been improved.
The context clear status in the text input area now supports persistent display. The displayed state automatically adapts based on the current **CB/CC mode** and whether the user is working with **Team Agent** or **Private Agent**.
The `/clear` command behavior remains unchanged.
## **Model updates**
The default model has been updated to **GLM-5.1**.
**Claude Sonnet 4.6** remains available and can still be selected manually.
This release also adds support for new models, including:
* **Qwen3.7-Max**
* **GPT-5.5**
* **Claude Opus 4.8**
These updates expand model availability and provide users with more options for different development scenarios.
## **DC Agent, Messaging, Skill Market, and Collaboration Experience Updates**
This release includes updates across Discussion, messaging, Skill Market, organization structure, agent capabilities, viewer support, and multiple interaction improvements across the product.
### **DC Agent** New
**Discussion** now supports **DC Agent**.
This gives teams an additional agent directly inside Discussion, making it easier to coordinate conversations, keep context together, and support group collaboration workflows.
### **Message queue mechanism** New
Messaging now includes a **queue mechanism**.
This improves message handling in scenarios where multiple actions or requests happen in sequence, making the overall interaction flow more stable and easier to follow.
### **Skill Market updates** Improve
The installed Skill list now includes an **Org Badge** indicator.
Skill-related services have also been significantly improved, with better stability for skill download and backup workflows. These updates make Skill usage and management more reliable across projects.
### **Organization structure and Contacts updates** Improve
This release includes several improvements to organization structure and contact management.
Fixed an issue where organization structure data was not written correctly in some registration scenarios. The department tree in **Contacts** now supports expand and collapse behavior, making organization browsing clearer. Organization nodes and external contact nodes now also include icons for better visual recognition.
### **Agent updates** Improve
Agent conversations now include **message timestamps** and **cross-date markers**.
These updates make date changes in conversations easier to recognize and help reduce confusion in time-sensitive agent interactions.
Agent capabilities have also been expanded with new workflow support. Agents can now create **webhook tasks**, enabling more flexible integrations and automation scenarios. Group chat management has also been expanded, with support for deleting group chats and updating group chat information.
This release also adds multimedia generation capabilities. Agents can now support both image and video generation workflows.
In addition, new model options have been added, model-related configuration issues have been fixed, and the model list order has been updated. Project creation now also supports improved multilingual behavior, with better support for internationalized experiences.
### **Viewer update** Improve
The **Viewer** area now supports previewing compressed archive files.
This expands in-product preview support and makes it easier to inspect packaged files without leaving the workspace.
### **Input box and message experience updates** Improve
The input and message experience has been improved across attachments, search, quoting, and long-message handling.
Attachments are now cached by tab and project, which helps prevent uploaded files from being lost when switching pages. Fixed an issue where pinyin-based user search in the input box could return unrelated members. Fixed a case where numbers could occasionally be lost when using **Sogou Input Method**. Quoted message line wrapping has also been improved.
Sent long messages now support automatic **expand / collapse** behavior, making lengthy content easier to read.
### **@ mention update** Improve
Fixed an occasional issue where the **@ mention** panel could appear even when no content had been entered after triggering **@**.
This makes the mention experience more stable and predictable.
### **Project name display update** Improve
Fixed an issue where project names could fail to display correctly when the **Owner** nickname was too long.
This improves readability in project-related views.
### **Project list updates** Improve
The **Project List** and **Chat List** now support automatic positioning.
When switching between the two tabs, the system now automatically associates the corresponding item and keeps it within the visible area. The pin and unpin prompt copy has also been updated in Chinese.
### **Members and agent name updates** Improve
The **Members** dialog has been improved so that it closes automatically after an invite is sent.
This release also improves the display and naming behavior of **Personal Agent**. Fixed an issue where the **Owner** could be missing from the Members list, and fixed a case where entering **@** could send directly in unintended scenarios.
### **A2A dialog update** Improve
The A2A dialog has been optimized for group owners.
Group owners no longer see the permission dropdown in this flow and now default to **Can Edit**.
### **Forwarding and avatar display updates** Improve
Fixed an issue affecting avatar and display name presentation in forwarded messages.
This makes forwarded content easier to read and improves message source recognition.
### **Conversation flow updates** Improve
The **MermaidDiagram** flowchart component now includes a close button in fullscreen mode.
This release also fixes and improves formatting behavior for user-sent rich content, making conversation rendering more consistent.
## **Team Plan, Organization, Agent, Viewer, and Interaction Updates**
This release includes updates to Team Plan billing and member management, organization structure handling, agent collaboration, archive preview, input behavior, project navigation, A2A interactions, and CBBot.
### **Team Plan updates** New
CodeBanana now converts **requests** into **usage**, with previously purchased entitlements migrated at a **10:1 ratio**.
For organizations that already have **Team Plan** enabled, adding seats and upgrading VM tiers can now be charged through the shared usage pool. A confirmation dialog has also been added to make these actions clearer before submission.
This release also includes pricing updates for selected plans.
### **Team Plan alerts** New
The Team Plan alert system has been expanded.
A reminder is now shown when remaining usage falls below **20%**. A synchronized warning is also displayed when the remaining limit balance falls below **\$10**.
These updates improve visibility before usage or balance is fully exhausted.
### **Team Plan member management** Improve
Both organization **Admins** and **Owners** can now update member quotas, VM configurations, and available model access.
Previously, these actions were limited to the **Owner** only. This change makes day-to-day Team Plan management more flexible for organizations with delegated admins.
### **Team Plan top-up backend** New
The recharge backend now supports **custom-day top-ups**.
This includes support for trial quota top-ups and provides more flexibility for temporary or time-based usage allocation.
### **Team Plan fixes and billing improvements** Improve
This release also includes a set of Team Plan fixes and billing-related improvements.
The add-seat and VM-upgrade flows now display the billing amount more clearly. Fixed an issue where used seat counts did not refresh automatically after adding or removing members in Team Plan. Fixed an issue where members could not be removed from the confirmation page when adding members. Fixed an issue where the model list could appear empty in group chats for non-Team Plan members.
In addition, fixed an issue with incorrect dates in exported **Usage** Excel files, and improved the billing page experience.
### **Organization structure update** Improve
Fixed an issue where organization structure data was not written correctly in some registration scenarios.
This improves data consistency during organization onboarding and setup.
### **Agent updates** Improve
Agent conversations now include message timestamps and cross-date indicators.
These updates make date changes inside conversations more visible and help reduce confusion in time-sensitive agent interactions.
This release also introduces the **Discussion Agent (Discussion Coordinator)**.
The Discussion Agent can summarize group content and answer questions based on discussion context. It can also call other agents through **A2A**, with the Discussion Agent acting as the coordination layer. The Discussion Agent is visible in both the **Members** list and the **@ mention** list, and it can be invoked directly through **@**.
### **Viewer update** Improve
The **Viewer** area now supports previewing compressed archive files.
This expands file preview support for common packaged project assets and documents.
### **Input box updates** Improve
The input experience has been improved across attachments, search, input methods, and quoting behavior.
Attachments are now cached by tab and project, which prevents uploaded files from being lost when switching pages. Fixed an issue where pinyin-based user search in the input box could return unrelated members. Fixed a case where numbers could be lost when using **Sogou Input Method** in the message input field. Quoted message line wrapping has also been improved.
### **@ mention update** Improve
Fixed an occasional issue where the **@ mention** panel could appear after sending **@** even when no content had been entered.
This makes the mention interaction more stable and predictable.
### **Project name display update** Improve
Fixed an issue where project names could fail to display correctly when the **Owner** nickname was too long.
This improves readability in project-related views.
### **GitLab import updates** Improve
GitLab and GitHub merged project names are now derived from the Git URL.
The **GitLab import** dialog has also been improved, including better content height handling and layout refinement.
### **Project list updates** Improve
The **Project List** and **Chat List** now support automatic positioning.
When switching between the two tabs, the system now automatically associates the corresponding item and keeps it within the visible area. The pin and unpin prompt copy has also been updated in Chinese.
### **Members and agent name updates** Improve
The **Members** dialog has been improved so that it closes automatically after an invite is sent.
This release also improves the display and naming behavior of **Personal Agent**. Fixed an issue where the **Owner** could be missing from the Members list, and fixed a case where entering **@** could send directly in unintended scenarios.
### **A2A dialog update** Improve
The A2A dialog has been optimized for group owners.
Group owners no longer see the permission dropdown in this flow and now default to **Can Edit**.
### **Conversation flow updates** Improve
The **MermaidDiagram** flowchart component now includes a close button in fullscreen mode.
This release also fixes and improves formatting behavior for user-sent **rich\_content** messages.
### **CBBot update** Improve
**CBBot** now supports creating cloud projects.
This expands project creation options for users working across local and cloud-based workflows.
## **Usage Alerts, Forwarding, and Collaboration Experience Updates**
This release introduces usage alerts and includes a broad set of updates across forwarding, A2A permissions, member management, mentions, quoting, input behavior, search, and avatar display.
### **Usage alerts** New
CodeBanana now supports usage alerts for both request-based and token-based limits.
Alerts now apply to free requests, included plan requests, Team Plan token limits, and shared token pools. A warning is shown when remaining usage falls below **20%**, while existing reminders for exhausted usage remain unchanged.
### **Forwarding updates** Improve
Forwarding now supports sending messages directly to contacts in **Contacts**, in addition to forwarding to projects.
This release also fixes an issue where nested forwarded messages containing images could display incorrectly. It also fixes control conflicts related to forwarded messages under shared-control scenarios.
### **Bulk group creation** Improve
Backend support is now available for bulk group chat creation, including multi-member group setup.
This improves efficiency for large-scale group creation and management workflows.
### **Personal Agent project protection** Improve
Projects associated with **Personal Agent** can no longer be deleted.
This helps prevent accidental deletion of default personal agent workspaces.
### **A2A permissions and guidance updates** Improve
A2A permissions and related behavior have been refined.
Agent permission guidance copy has been updated and is now supported across **English**, **Simplified Chinese**, and **Traditional Chinese**.
### **Member management and permission dialog improvements** Improve
The member and permission management experience has been simplified.
The permission dropdown and the three-dot action menu have been removed, and related actions are now managed through a unified settings entry point. Member sorting has also been updated to:
**coding\_agent → Owner → Admin → Member**
This release also fixes an issue where permission-related dialogs could display at the wrong visual layer. Under shared-control mode, both **Admin** and **Owner** can now forward messages to **@agent** in **Discussion** without checking presenter status.
### **@ mention improvements** Improve
The **@ mention** experience has been improved.
The mention dialog now provides search result counts and shows **Recent files** by default. This release also fixes an issue where agent names could occasionally be truncated when using **@agent**.
### **Copy and paste improvements** Improve
Copy and paste behavior has been improved for messages with attachments.
Previously, copied messages with attachments could display only the attachment itself. This has been fixed so the actual filename is now shown correctly.
### **Quote improvements** Improve
The quoting experience has been improved across both layout and display.
Attachments now appear above text, and text appears below. Multiple images are now shown as thumbnails. Video and file attachments now display type labels such as **\[Video] filename**, and sender names are now displayed in a more consistent format.
### **Input box improvements** Improve
The input box behavior has been improved after sending long messages.
This release fixes an issue where the input box could remain stuck at its maximum height after sending a message. After long text is sent, the input box now returns to its default height correctly.
### **Search bar redesign** Improve
The search bar has been redesigned for better usability.
It is now centered and uses a wider layout to better support both file search and content search.
### **Avatar display improvements** Improve
Default avatar display logic has been improved.
If a user name contains Chinese characters, the default avatar now uses the last two Chinese characters. If no Chinese characters are present, the first four characters are used instead. Group avatars in **chatList** have also been updated.
## **Team Plan, Cross-Project Agents, and Onboarding Updates**
This release introduces a redesigned Team Plan, cross-project agent collaboration, model access controls, onboarding improvements, and a set of usability updates across scheduling, skills, contacts, and project setup.
### **New Team Plan** New
CodeBanana now supports a redesigned **Team Plan**.
The new Team Plan supports customizable seat counts starting from 1 seat, along with a shared **Token pool** with no minimum threshold. Three VM tiers are available at **\$45 / \$65 / \$105 per seat per month**, with both monthly and annual billing supported. Annual billing is available at **20% off**.
Admins can manage seats, token top-ups, member limits, renewals, and related usage settings from the **Usage** page. Enterprise invoice payment and backend balance top-up are also supported.
### **A2A — Cross-project agent collaboration** Improve
CodeBanana now supports **A2A**, enabling agents to collaborate across projects.
Each project agent is named after its project. The **Members** list now supports adding agents authorized by other users, as well as agents owned by the current user. Owners can assign one of three permission levels to an agent: **Can Edit**, **Can Read**, or **Disabled**.
Each user also has a preconfigured **Personal Agent**, which is automatically included when invited into a group. In addition, agents can send messages to **Discussion** or private contacts, query contacts, and **Team Agent** can read discussion history.
These updates expand how agents participate in collaboration across projects, chats, and teams.
### **Model access controls** New
Admins can now define available model ranges for members at the individual, bulk, or full-selection level.
Once a member is restricted, the restriction applies to all projects owned by that member. This gives organizations more centralized control over model usage across teams.
### **Onboarding improvements** Improve
This release also expands onboarding across projects and organization setup.
Each project now includes the **CB-FAQ Skill** by default, with synchronized product updates. The file tree now includes built-in **User Guides**, including role-based use cases, and **Agent bootstrap** can jump directly to the relevant guidance.
Organization admins also now have access to a dedicated **Org Admin Workspace**. This workspace includes seven core capabilities, such as organization structure setup, department management, member and invitation management, AI usage visibility, and AI maturity recommendations.
New users are assigned this workspace by default. Existing users can access it from **Template → Data & Analyst**.
### **Experience improvements** Improve
This release also includes multiple experience and interface updates across the product.
Updates include:
* Upgraded **@ mention** dialog styling, now with search suggestions
* A new **Cron Job** page with execution details such as task duration
* **Private Agent** renamed to **Private Ask**
* Improved **Git Commit / History** dialog
* Updated **Create Project** templates and dialog experience
* **Skill** version management now supported with automatic version increments, and author information is now displayed in the list
* **Contacts** now supports directly adding external friends
* Organization friend request history now supports categorized views
These updates improve clarity, discoverability, and day-to-day workflow efficiency across the product.
## **Admin, Agent, Skill, and Dashboard Updates**
This release includes updates to group admin controls, agent collaboration flows, skill versioning, scheduled tasks, the AI-Native Index Dashboard, contacts, and several core product interfaces.
### **Group admin controls** New
CodeBanana now supports expanded group admin capabilities.
The first five members added to a group are assigned as default admins. Admins can also compete for control equally even when there is no active agent request. In addition, group owners can now configure admins directly from the top-right corner of the member dialog, with support for up to five admins per group.
These updates make group-level collaboration and permission coordination more flexible in multi-user project scenarios.
### **Add Agent** New
The **Add Agent** flow now supports search.
Users can now search for agents directly when adding them, which makes it easier to find and assign the right agent in larger workspaces or more complex project setups.
### **Agent collaboration updates** Improve
Several agent collaboration capabilities have been expanded in this release.
Agents can now send messages to **Discussion** and also send messages directly to individual contacts through private chat. Agents can also query contact members. In addition, **Team Agent** now supports reading chat history from **Discussion**.
These updates improve context sharing between agents, group discussions, and direct communication flows.
### **Skill updates** Improve
Skills now support version numbers and incremental versioning.
When a skill is updated, its status will also be synchronized accordingly. The skill list now displays author information as well.
These changes make skill management more transparent and improve traceability across updates and reuse.
### **Scheduled task improvements** Improve
Scheduled tasks now provide additional execution details, including time consumption.
This makes it easier to review task performance and understand execution behavior over time.
### **AI-Native Index Dashboard updates** Improve
The **AI-Native Index Dashboard** has been expanded with additional organization and department visibility.
Organization admins and owners can now view department and organization information. Department heads can also view department-level usage data.
These updates improve visibility for organizational management and support more role-based oversight.
### **Contacts updates** Improve
The **Contacts** section now supports directly adding external friends.
Organization friend request history also now supports categorized views, making it easier to review and manage different types of contact requests.
### **Interface improvements** Improve
This release also includes a set of interface and interaction updates across key product areas.
Updates include:
* Improvements to the **Cron Job** page
* **Private Agent** renamed to **Private Ask**
* UI improvements to the **Git Commit / History** dialog
* Updated project creation templates and improved dialog experience
These updates continue to improve clarity, consistency, and usability across the product.
## **Documentation, Contacts, and Viewer Updates**
This update includes improvements to the documentation, contact management, invitation flow, and file preview capabilities.
### **Documentation center** New
CodeBanana now includes a dedicated **Documentation** tab in both the homepage and Personal Center.
The documentation center now supports question search and synchronized changelog updates, making it easier for users to find product information and keep up with the latest changes in one place.
### **Default project** New
Default projects now include feature guidance files.
This helps users understand key product capabilities more quickly after entering a project and provides a clearer starting point for exploration and onboarding.
### **Contacts** New
The **Contacts** section now includes both internal organization contacts and external contacts, along with related contact guidance.
These updates improve contact visibility and make it easier for users to understand who is available within the organization and who is managed as an external contact.
### **Invite dialog** New
The invite dialog now supports inviting both internal and external contacts directly.
This removes the previous restriction and makes collaboration setup more flexible across different communication and team scenarios.
### **Viewer preview improvements** Improve
The Viewer preview area has been improved for a wider range of document types.
Preview support and viewing experience have been optimized for files including:
* DOC / DOCX
* PDF
* CSV
* Excel
These updates improve document readability and make in-product file review more consistent across common file formats.
## **New user onboarding and organization management**
CodeBanana now supports onboarding for new users, along with organization structure and permission management. These updates are now live in production at [www.codebanana.com](http://www.codebanana.com).
This release covers onboarding flows for individual users, team users, and enterprise users, helping each user type complete setup more efficiently and get started in the right workflow. All onboarding steps can be skipped and completed later through Contacts.
### **Personal onboarding** New
Individual users can now select and submit their intended use case during onboarding.
This helps CodeBanana to provide a more relevant initial experience.
### **Team and enterprise onboarding** New
CodeBanana now supports a more complete onboarding flow for team and enterprise users.
When an organization owner signs up, a default organization is created automatically, and the owner can rename the organization during onboarding. Teams can also upload organization structure data through an excel file, including department and member information, to support bulk invitation, registration, and team structure setup.
Organization members can directly search for and join the target organization during onboarding. All onboarding actions are optional and can be skipped, with further updates and edits available later in Contacts under Studio page.
### **Organization structure upload and management** New
Organization structure upload and management now support a more complete set of admin and member management workflows.
Organization owners can now assign multiple organization admins. Organization owners and admins can edit member roles and department information, and create both top-department and sub-departments.
The following management capabilities are also now supported:
* Invite organization members individually or in bulk
* View invitation records and invitation progress
* Review and approve organization join requests
* Upload and update organization structure data in bulk
* Remove organization members
* Transfer projects from departing members and choose the receiving owner
These updates make it easier for teams to complete both initial organization setup and ongoing admin work in one place.
## Team Plan, Agent, and Workspace Experience Updates
This production update includes three main areas, covering multiple features and product improvements:
### **Team Plan updates**
CodeBanana now includes a more complete Team Plan workflow, including purchase, member management, and usage limit controls.
These updates make it easier for teams to manage plan access, assign members, and control quotas more clearly within the workspace.
### **Agent capability upgrades**
This release expands CodeBanana’s agent capabilities across project setup, configuration, and external integrations.
Updates include:
* Claude Code Agent support
* Agent Config settings support
* Skills Market support
* IM bot integration
* Agent template selection when creating a new project
These changes give teams more flexibility when configuring agents and reusing workflows across different project scenarios.
### **UI, project, and developer experience improvements**
This release also includes updates to the product interface, project file management, and developer tooling organization.
These improvements are designed to make navigation clearer, streamline file operations, and provide a more integrated experience for technical users working inside projects.
## **Coming next this week**
The following updates are planned next:
* **AI-Native Index Dashboard**
* Enterprise onboarding, along with organization structure upload and management
* Continued UI and interaction improvements, including project and file preview/editing experience, as well as agent chat and interaction experience
### Major Feature Updates
### New Features
* **Sharing System Launch:**
* **Project Sharing:** Share projects via link with team members and collaborators.
* **Enhanced Data Analytics:**
* **Analytics Dashboard Redesign:** More powerful data visualization and analysis capabilities.
* **Feature Usage Statistics:** Track usage patterns across all features.
* **Feature Rankings:** See the most popular features at a glance.
* **Voice Functionality:**
* **Voice Recording:** Record audio and convert to text automatically.
* **Technical Support System:**
* **Tech Support:** Professional technical assistance service.
* **Chat Enhancements:**
* **@Mention Notification Cards:** Dedicated display for mention notifications.
* **Chat List Search:** Search contacts and organization members.
* **Scroll to Bottom Button:** Quickly jump to the latest messages.
* **Automatic URL Recognition:** Links are automatically clickable.
* **Membership System Upgrade:**
* **VM Specification Selector:** Customize virtual machine configurations.
* **Membership Plan Comparison:** Feature comparison table for all plans.
* **Free Tier Optimization:** Clearer display of free plan features.
### Bug Fixes
* Fixed chat draft switching timing issues.
* Fixed scroll to bottom button positioning.
* Resolved Git status reload problems.
* Fixed file context display issues.
* Corrected message forwarding handling.
### New & Improved
* **Analytics Dashboard 2.0:** Comprehensive upgrade to the analysis dashboard, featuring a new interactive click-tracking board for deeper insights.
* **Enhanced Share Viewer:** Improved the user experience and reliability of the file sharing viewer.
### Fixes
* **Message Forwarding:** Resolved an issue where message forwarding was malfunctioning.
* **Localization:** Fixed internationalization (i18n) issues on registration-related tooltips.
* **UI Navigation:** Fixed the positioning logic of the "Scroll to Bottom" button for smoother chat navigation.
### Changes
* **Terminology Update:** Renamed "Assistant" to **"Tech Support"** across the entire platform to reflect a more professional service model.
### Improvements
* **Analytics:** Added date range selection to the analytics dashboard and optimized the data export function.
* **Voice Transcription:**
* Disabled the "Cancel" button during active transcription to prevent accidental data loss.
* Resource optimization: Canceling a recording no longer triggers an unnecessary API call.
### New Features
* **Telemetry & Analytics:** Implemented comprehensive tracking for:
* Viewer operations.
* Service and Terminal button usage.
* Git history and commit actions.
* Navigation events and custom notifications.
* **Chat Enhancements:**
* **User Profile Cards:** New design including privacy protection measures.
* **Draft Persistence:** Unsent messages are now automatically saved and restored.
### Security & Infrastructure
* **Email Validation:** Enforced strict email address validation across all mail services.
### Improvements & Fixes
* **Sharing Logic:** Restricted the "Share" button functionality to static previews only.
* **File Handling:**
* **Git Status:** Added support for file paths containing special characters.
* **PDF Viewer:** Fixed URL encoding issues for filenames with special characters.
* **Excel Editor:** Added proper display support for Chinese characters.
* **Downloads:** Fixed an issue causing binary file downloads to fail.
### UX Improvements
* **Feature Visibility:** The **"Request Assistance"** feature is now more prominently displayed in the UI for easier access.
### UI/UX Polish
* Refined indicators for Group Owners.
* Optimized preview states and message bubble styling.
### Organization & Deployment
* **Group Management:** Enhanced functionality for member management and group notifications.
* **Deployment:** Updated Tencent Cloud configurations and port settings.
### Fixes
* Resolved rendering issues with dropdown menus.
* Fixed property naming errors affecting frontend stability.
# Managing your Account
Source: https://docs.codebanana.com/documentation/account-and-subscriptions/account
Manage your CodeBanana profile, avatar, usage, app defaults, organization contacts, external contacts, external friends, invitation history, and personal account settings.
Your account settings bring together everything that is personal to you, from your name and profile image to how you send messages in chats. Use this page to keep your details accurate and safe.
## **Accessing your account settings**
You can manage all your personal account options from the profile menu.
**To open your account settings:**
1. Enter the main page of CodeBanana.
2. Click the profile picture in the upper right corner.
### Change profile picture
Click on your profile picture to access the profile picture editing page. Here you can upload an image to change your profile picture.
The profile picture must meet the **following requirements**:
1. Only JPG, PNG, GIF, and WebP formats are supported.
2. Less than 5MB.
![Header]()
### Usage
Click the "Usage" button in your personal center to view your usage details. Overall billing is divided into **two types**:
1. **Package-based requests:**
* Different membership levels—Free, Personal, and Professional—offer 30, 300, and 1000 requests respectively.
* The Personal and Professional packages allow you to choose different virtual machine options based on your service scale; billing will be based on the virtual machine service used.
2. **Pay-as-you-go**
* Once you've used up your package-based requests, you can enable pay-as-you-go billing on the Usage page, setting a limit. This will automatically trigger a Stripe billing process for top-up.
![Usage]()
### Contacts
You can view your organization application records, organization members, and external contacts.
The entry to **Contacts** depends on whether **Standard Mode** is enabled.
**Standard Mode On**
When **Standard Mode** is enabled, **Contacts** is available directly from the left navigation bar.
![0715 Contacts Standard]()
**Standard Mode Off (Classic Layout)**
When **Standard Mode** is disabled, open the profile menu in the upper-right corner, then click **Contacts**.
![0715 Contactold]()
#### Organization Contacts
Members who have joined the same organization. You can:
* Add them directly to projects without sending email invitations
* Browse and search all organization members in one place
* View each member’s role (Owner / Admin / Member)
#### External Contacts
Collaborators invited by email who are not part of the organization. They become external contacts after accepting an invitation sent either from a project or through **Contacts → External Contacts → Add External Contact**. You can:
* Invite them to projects via email without adding them to the organization
* Find and reuse them when inviting to other projects
* Distinguish them visually from internal members to avoid confusion
* View invitation records and statuses in **History**
| | **Organization Contacts** | **External Contacts** |
| ------------------- | ----------------------------------------- | ------------------------------ |
| How they join | Join the organization directly | Invited to a project via email |
| Add to project | Can be added directly (no email required) | Requires email invitation |
| Organization member | Yes | No |
* **Admin and owner** can manage the organization structure and members, like inviting members and setting admins.
See more details in Organization Page.
* Here offers a convenient search function for quickly finding people, and you can also click the chat button next to them for quick private chats. At this point, you will be redirected to the chat option, and all private chat history will be saved there.
![Contactschatwithmember]()
### **Language Setting**
CodeBanana supports multiple interface languages, allowing you to work in the language you're most comfortable with.
Currently supported languages:
* **English**
* **简体中文 (Chinese)**
To change the interface language:
1. Open the profile menu.
2. Click **Language**.
3. Select your preferred language. The interface updates immediately after the new language is selected.
Your language preference is saved to your account and will be used the next time you sign in.
![0715 Language]()
### **Standard Mode**
CodeBanana provides two workspace layouts for different ways of working.
To switch layouts:
1. Open the profile menu.
2. Turn **Standard Mode** on or off.
![0715 Switch Standardmode]()
#### **Standard Mode On (Default)**
When **Standard Mode** is enabled, CodeBanana uses the **new workspace layout**. This layout is optimized for **AI-first collaboration**, making conversations with agents more prominent while keeping your files and workspace visible.
It is recommended for users who frequently work with **Team Agent** or **Private Ask**. Features include:
* More space for agent conversations
* Better visibility of agent status and execution progress
* Optimized for AI-assisted coding and collaborative workflows
![0715 Standardmodeon]()
#### **Standard Mode Off (Classic Layout)**
When **Standard Mode** is disabled, CodeBanana switches to the **classic workspace layout**. The classic layout allocates more space to the editor and file viewer, making it easier to focus on project content. It is recommended for users who primarily read, edit, or review files.
Features include:
* Larger editor and preview area
* More space for project content
* Reduced emphasis on the chat panel for a more document-focused experience
![0715 Standardmodeoff]()
You can switch between the two layouts at any time. This setting only changes the workspace layout and does not affect your projects, conversations, or files.
# Billing & Plan
Source: https://docs.codebanana.com/documentation/account-and-subscriptions/billing-plan
Compare CodeBanana Free, Personal, and Professional plans, including included token quota, VM tiers, monthly or yearly billing, token top-up, usage alerts for individual plans, and how token-based billing works.
## **Choosing your plan**
CodeBanana offers flexible plans for different levels of usage, from free trial access to heavier personal workloads.
### **Free**
Free gives you full access to CodeBanana with a one-time trial quota.
* **\$5 free token quota**
* Valid during the **registration month**
* **20 MB** cloud storage
* No feature gating
Free plan token quota is issued once only. It expires at the end of the registration month and is not reissued.
### **Personal**
Personal is designed for individual developers who need a dedicated environment and a higher included token quota.
* **\$25 included token quota / month**
* **1 GB** cloud storage
* Multiple VM tiers:
* **2 vCPU / 4 GB RAM**
* **4 vCPU / 8 GB RAM**
* **8 vCPU / 16 GB RAM**
Personal plan pricing is made up of two parts: the VM fee and the included monthly token quota fee.
### **Professional**
Professional is designed for heavier individual workloads and larger projects.
* **\$130 included token quota / month**
* **5 GB** cloud storage
* Available VM tiers:
* **4 vCPU / 8 GB RAM**
* **8 vCPU / 16 GB RAM**
Professional plan pricing also combines the VM fee and the included monthly token quota fee.
View pricing and choose the plan that fits your needs.
***
## **Purchasing a plan**
Once you have selected the plan that fits your needs, you can upgrade from the pricing page.
**Have a promo code?** Apply it **before** proceeding to checkout, as it will not take effect automatically afterward.
**To purchase a plan:**
1. Click the **upgrade button** at the top right of your workspace
2. Choose a plan
3. Select a VM tier
4. Select a billing cycle:
* **Monthly** — pay month to month
* **Yearly** — save with an annual discount
5. Click **Upgrade** to continue to checkout
![0520 Pricing]()
### **Yearly pricing**
Yearly billing applies a discount to the combined annual cost of:
* The VM fee
* The included token quota fee
This means yearly pricing is calculated from the full plan value rather than from the VM fee alone.
***
## **Token Quota and Billing**
### **What are Tokens**
**Tokens** measure the actual amount of model computation used during a task.
This includes:
* **Input** — prompts, context, files, and instructions
* **Output** — agent responses, generated code, and plans
More complex tasks, larger context windows, and more advanced models typically consume more tokens.
### **Included Token Quota**
Each plan includes a built-in token quota:
* **Free** — **\$5** one-time token quota
* **Personal** — **\$25 / month**
* **Professional** — **\$130 / month**
This included quota is part of the plan itself.
### **Token Top-up**
If your included token quota is exhausted, you can continue using CodeBanana through **On-demand Usage (Top-up)**.
Top-up is available for individual plans and lets you add token balance as needed without changing your current subscription.
Team Plan does not use personal top-up. Team usage is handled through the shared team token pool.
### **How Billing Works**
For individual plans, billing follows this structure:
1. Your subscription price includes:
* A **VM fee**
* An **included token quota fee**
2. Included token quota resets on your **subscription renewal date**
3. Unused included quota does **not** roll over into the next cycle
4. If your included quota is exhausted, you can continue through **Top-up**
### **Usage Alerts**
CodeBanana provides usage alerts to help individual users track remaining quota.
* A reminder is shown when the remaining token balance in your plan falls below **10%**
***Reminder text***
> You're running low on tokens — \$ left in your plan. Top up or upgrade now to keep your work uninterrupted.
Usage alerts apply to:
* Free plan token quota
* Included token quota in paid individual plans
If you upgrade from Free to a paid individual plan while your remaining balance is already low, CodeBanana will not immediately send the same low-token reminder again.
***
## **FAQ**
Free plan includes a one-time **\$5 token quota**, **20 MB** cloud storage, and access to all core product features without feature gating.
No. Free plan quota is issued once during the registration month. It expires at the end of that month and is not reissued.
If you are using an individual paid plan, you can continue through **On-demand Usage (Top-up)** without changing your subscription.
Yes. Individual paid plans support token top-up after the included quota is used up.
Yearly pricing is based on the annual combined value of the VM fee and the included token quota fee, with an annual discount applied.
CodeBanana shows a reminder when the remaining token balance in your plan falls below **10%**.
Reminder text:
> You're running low on tokens — \$ left in your plan. Top up or upgrade now to keep your work uninterrupted.
No. Unused included token quota does not roll over into the next billing cycle.
# Introduction
Source: https://docs.codebanana.com/documentation/cbbot/introduction
Set up CodeBanana Bot, the local desktop version of CodeBanana for macOS and Windows, and understand account sync, cloud and local projects, local files, terminal commands, app control, browser automation, screen understanding, and bot-only skills.
## What is CodeBanana Bot
Unlike the web version, CodeBanana Bot operates within your local environment, allowing the agent to interact with your **files, applications, browser, and system**, extending AI capabilities beyond the cloud into your personal workspace.
## Web vs CodeBanana Bot
CodeBanana Bot and the Web version are **two forms of the same product**, sharing the same account system and synchronized data.
| Dimension | Web | CBBot |
| ------------------ | ------------------------------------------------- | ---------------------------------- |
| Runtime | Browser-based | Local desktop application |
| Core features | Full product capabilities (projects, chats, orgs) | Same core capabilities |
| Agent scope | Cloud projects & environments | Local system, files, apps, browser |
| Account | Shared | Shared |
| Project visibility | Cloud + local (when client is online) | Cloud + local |
| Best for | Collaboration & accessibility | Deep local automation |
> Both interfaces display the same project list. Cloud projects are always accessible, while local projects are available when the client is online.
***
## Installation
CodeBanana Bot supports both macOS and Windows.
| Platform | Support |
| -------- | ------- |
| macOS | ✅ |
| Windows | ✅ |
### Steps
1. Download the installer from the CodeBanana website
2. Install:
* macOS: Open `.dmg` and drag to Applications
* Windows: Run `.exe` installer
3. Launch CodeBanana Bot and log in with your account
***
## Login & Sync
### Account
* Use the same CodeBanana account as Web
* Supports email and third-party login
* Can stay logged in on both Web and CBBot
### Data Sync
| Data | Behavior |
| -------------- | --------------------------------------------- |
| Account info | Synced in real time |
| Billing & Plan | Shared across platforms |
| Organization | Shared members & permissions |
| Team Plan | Fully synced |
| Cloud projects | Fully synced |
| Local projects | Synced; shown as offline if client is offline |
| Chat history | Synced across devices |
> Project configurations (Basic Info, Skills, Cron Jobs) are project-specific but consistent across platforms.
***
## Local Agent Capabilities
CodeBanana Bot enables agents to operate directly on your **local machine**.
### System Control
* Read, create, modify, delete files
* Move and batch process files
* Execute terminal commands and scripts
* Access system-level information
### Application Control
* Open local applications
* Simulate user interactions (click, type, etc.)
### Browser Automation
* Navigate web pages
* Fill forms and extract data
* Simulate full browsing workflows
### Screen Understanding
* Capture screen content
* Analyze UI states
* Validate operation results
***
## Typical Use Cases
| Scenario | Workflow |
| ----------------- | ------------------------------------------ |
| File organization | Scan → classify → move files |
| Form automation | Open browser → fill → submit |
| Reporting | Capture screen → analyze → generate report |
| File monitoring | Trigger workflows via File Watch |
| App automation | Open app → execute actions |
***
## Bot-only Features
### Bot-exclusive Skills
CodeBanana Bot supports **Bot-only Skills**, which include local system operations.
| Type | Availability |
| --------------- | -------------------- |
| Standard Skills | Web + Bot |
| Bot-only Skills | Bot only (`Bot` tag) |
***
### Cron Job Differences
| Capability | Web | CBBot |
| -------------------- | ------------------ | --------------------- |
| Scheduled tasks | ✅ | ✅ |
| System notifications | ❌ | ✅ |
| File Watch scope | Project files only | Full local filesystem |
***
## Project Status
| Type | Location | Availability |
| ------------- | ------------- | ------------------------------ |
| Cloud Project | Server | Always available |
| Local Project | Local machine | Offline when client is offline |
```text theme={null}
Projects
├── Cloud Project A (always available)
├── Cloud Project B (always available)
├── Local Project C (online available)
└── Local Project D (online available)
```
## Permissions
Permission setup differs by system.
**Path:** CBBot → Settings → Permissions
| Setting | Required | Description | macOS | Windows |
| -------------------- | ----------- | -------------------------------------------------------------------------------------------------- | ---------------------------- | -------------- |
| **Keep Awake** | Optional | Keeps the system active when the lid is closed or the screen is off (requires power connection) | ✅ Supported | ✅ Supported |
| **Accessibility** | Recommended | Enables UI automation, including controlling system interfaces and simulating keyboard/mouse input | ✅ Manual permission required | — Not required |
| **Full Disk Access** | Recommended | Allows reading and writing files across the entire system | ✅ Manual permission required | — Not required |
| **Screen Recording** | Optional | Enables screen capture and recording (including system audio) | ✅ Manual permission required | — Not required |
> Accessibility and Full Disk Access are critical for core functionality. Some changes require restarting the app.
# Organization
Source: https://docs.codebanana.com/documentation/teams/org
Manage CodeBanana organizations, switching, join requests, owned and work organizations, members, roles, permissions, departments, organization contacts, invitations, application history, batch import, export files, and member removal.
The way you switch organizations or join a new one depends on whether **Standard Mode** is enabled.
#### **Standard Mode On**
When **Standard Mode** is enabled, click the **organization selector** in the lower-left corner.
From there, you can:
* Switch between organizations
* Apply to join a new organization
![0715 Joinorgnew]()
#### **Standard Mode Off (Classic Layout)**
When **Standard Mode** is disabled, click the **organization selector** in the upper-left corner.
From there, you can:
* Switch between organizations
* Apply to join a new organization
![0715 Joinorgnew]()
#### **Joining an Organization**
To join an existing organization:
1. Open the organization selector.
2. Click **Join Organization**.
3. Search for the organization by name.
4. Submit your join request.
5. After approval by the organization **Owner** or **Admin**, you can switch to that organization from the organization selector.
#### **Create Organization**
* A default organization is **automatically created** when you register (via email or third-party login)
* The organization name is initially based on your email and can be renamed by the owner
* Each user can have **only one owned organization**
#### **Organization Types**
CodeBanana distinguishes between two types of organizations:
**Own Organization**
* The organization you created and fully control
* You are the **Owner**
* Each user can have only one
**Work Organization**
* Organizations you join as a collaborator
* You can be part of multiple work organizations
#### **Key behavior**
* You can only be in **one organization at a time**
* Switching organizations means switching your entire workspace:
* Projects
* Files
* Contacts
#### **Members and Roles**
Organizations manage collaboration through roles:
* **Owner**
* Full control over the organization
* Manages members, permissions, and billing
* **Member**
* Can access AI resources and participate in projects
#### **Join mechanism**
* Users must **actively apply** to join an organization
* Search by organization name and submit a request
* Access is granted after admin approval
![Joinorg]()
#### **Leaving an organization**
* Users can switch between organizations if they belong to multiple
* Full removal currently requires action from the **organization owner**
* Self-service ***leave organization*** is not yet available (to prevent accidental data loss)
#### **Organization & Projects**
Projects are tightly bound to organizations:
**Strong binding**
* Every project must belong to a specific organization
* Projects cannot exist outside an organization
**Data isolation**
* Projects in one organization are **not visible** to members of another
* Switching organizations refreshes your entire project list
**External collaboration (special case)**
* If an external user is invited to a project:
* They gain access to the **project only**, not the organization
* Currently, the project may appear across their organization views
* Future updates will place such projects under the user’s personal organization
### Organization Management
**Admin and owner** can manage the organization structure and members through **Contacts page**.
#### **Overview**
The Contacts module serves as the central place for:
* Organization structure management
* Member management and permissions
* Bulk operations for scaling teams
![Org Contacts]()
#### **Invitations**
Members can be added in two ways:
* **Invite Members (Manual)**
![Clickavatar]()
* Invite users individually via the **Invite Members** button
![Clickinvitebotton]()
![Clickinvitebotton]()
* Enter the email address and select a department
![Individualinvite]()
* **Batch Import**
![Clickavatar]()
![Toprighticon]()
* Click Batch Import
* Download two templates
* Edit these templates under guidance
![Download Template]()
#### **Application History**
Application History shows all organization-related requests and invitations in one place.
It helps users track applications they submitted, invitations they sent, and requests they received from others.
* **Requests Sent**
* View the requests you submitted to join other organizations.
* You can check the status of each request, such as pending or approved.
* **Applications Sent**
* View the invitations you sent to others to join your organization.
* You can track the status of each invitation and cancel invitations that are still pending.
* **Requests Received**
* View the join requests submitted by other users to your organization.
* You can process incoming requests or review their previous status and handling results.
#### **Export files**
Exported files can be edited and re-uploaded:
* **Export Structure** — export department hierarchy
* **Export Members** — export member list
#### **Department Management**
**Admins and owner** can manage organizational structure:
* Create top-departments
* Add sub-departments under existing ones (**Add Sub-dept**)
* Edit department information
![Departmentmanage]()
* Delete departments
* Update member name, role, and assigned department
![Editmemberifo]()
* Remove Member
* Requires transferring all projects owned by the member to another user
* All organization permissions are revoked immediately upon removal
![Removemember]()
# Team Plan
Source: https://docs.codebanana.com/documentation/teams/teamplan
Configure CodeBanana Team Plan for organizations with customizable seats, shared Budget Balance, seat-based VM configurations, flexible validity periods, usage alerts, billing modes, member limits, batch member management, model templates, model usage limits, renewals, seat management, and organization-level billing records.
## **What is Team Plan**
Team Plan is a flexible organization subscription that centralizes seats, compute resources, AI usage, and administration in CodeBanana.
It allows organizations to manage:
* Customizable seat quantities
* Seat-specific VM configurations
* A shared **Budget Balance**
* Member on-demand limits
* Available models and reusable model templates
* Organization-level billing and usage records
Unlike individual plans, Team Plan AI usage is funded through a shared organization balance rather than a separate included token quota for each seat.
***
## **Get Team Plan**
Team Plan can be purchased from **Pricing → Team Plan**.
![0728 Teambuy 1]()
### **Seats and Budget Balance**
Seats and AI usage balance are purchased separately.
* Seat quantity can be selected based on organization needs
* A minimum of **1 seat** is required
* Each seat costs **\$20 / month**
* AI usage is deducted from the organization’s shared **Budget Balance**
* Budget Balance starts from **\$100**
* Budget Balance does not expire and can be topped up later
### **Validity period**
Team Plan supports multiple validity periods:
* **1 month**
* **3 months**
* **6 months**
* **12 months**
Available discounts may vary depending on the selected validity period.
### **VM configurations**
Seats are associated with specific VM configurations.
Available VM tiers include:
* **2 vCPU / 4 GB RAM**
* **4 vCPU / 8 GB RAM**
* **8 vCPU / 16 GB RAM**
The available seat quantity is tracked separately for each VM configuration.
***
## **Usage Entry**
After purchasing Team Plan, organization Owners and Admins can manage seats, Budget Balance, usage, and members from **Avatar → Usage**.
The Usage page provides access to:
* Seat usage and availability
* Shared Budget Balance and AI consumption
* Member-level limits and usage
* Model permissions
* Team Plan renewal
* Additional seat purchases
* Budget Balance top-up
* Seat and balance records
![0418 Teamplan Entry]()
***
## **Usage Management**
The **Usage** page provides organization-level management for Owners and Admins, while allowing members to view their own available usage and VM resources.
![0418 Teamplan Usermanage]()
### **Usage Alerts**
CodeBanana provides Team Plan usage alerts for member on-demand limits and the shared organization Budget Balance.
**Member on-demand limit**
* A reminder is shown when a member’s remaining on-demand limit is **10% or less and no more than \$10**
* Each user receives this reminder at most **once per day**
* Reminders are evaluated independently for each user
* The user is asked to contact the organization Owner or Admin before usage is interrupted
**Reminder text**
> You're almost at your on-demand limit (% used). Reach out to your organization owner or admin to increase your limit before usage is interrupted.
**Shared organization Budget Balance**
* A reminder is shown when the remaining Budget Balance falls below **10%**
* The reminder is intended for the organization Owner or Admin so the balance can be topped up in time
**Reminder text**
> Your organization's shared token budget is almost depleted — \$ left. Please contact your organization owner or admin to top up soon and keep your team running smoothly.
No reminder is shown when a member’s personal plan quota is exhausted while the organization still has available shared Budget Balance.
### **Team usage (Admin view)**
Organization Owners and Admins can view:
* Seats used and available
* Seat availability by VM configuration
* Shared Budget Balance
* Organization-wide AI usage
* Member usage and on-demand limits
* Seat records
* Budget transaction history
They can also:
* Add seats
* Top up Budget Balance
* Renew Team Plan
* Open **Members & Quotas**
* Review seat and balance changes
### **Personal usage (Member view)**
Each Team Plan member can view:
* Current token usage
* Available on-demand limit
* Current VM configuration
* Limit reset time
* Team Plan expiration information available to them
For Team Plan members, AI usage is deducted from the organization’s shared Budget Balance, subject to the member’s configured on-demand limit.
***
## **Admin Controls**
Organization Owners and Admins can manage Team Plan configuration, members, billing behavior, models, limits, and VM resources.
### **Add members to Team Plan**
Click the avatar in the top-right corner.
Select **Usage** from the dropdown to enter the usage management page.
![Teamplanstep1]()
Switch to the **Members & Quotas** tab.
Click **Add Members** to open the member selection panel.
![Teamplanstep3]()
Select members and configure their seats, limits, and model permissions before confirming.
When adding members, admins can:
* Filter organization members by department
* View available seats for each VM configuration
* Assign members to available seats
* Configure on-demand token limits
* Apply model permissions or model templates
* Adjust individual members separately when needed
Members must already belong to the organization before they can be added to Team Plan.
If the required VM seat is unavailable, CodeBanana can direct the admin to purchase an additional seat. Payment can use Budget Balance when sufficient, or continue through Stripe.
### **Billing mode**
Billing Mode determines whose usage identity is used when AI costs are attributed inside the organization.
Organization Owners and Admins can change the organization’s Billing Mode.
**Where to configure**
1. Click your **avatar** in the top-right corner
2. Select **Billing Mode**
3. Choose a billing mode
4. Click **Save**
![0715 Billingmode]()
**Available billing modes**
* **Owner Mode** — Usage is attributed to the project Owner
* **Presenter Mode** — Usage is attributed to the current presenter based on their actual activity
The selected Billing Mode applies to the entire organization. The AI-Native Index Dashboard also displays usage statistics based on the selected mode.
### **Model templates**
Team Plan admins can create reusable model configurations for different teams or types of work.
A model template can define:
* Available models
* Model display order
* Default model
* Whether **Auto mode** is used as the default
Examples include:
* Product and engineering template
* Marketing and operations template
* Cost-controlled template
After a template is created, admins can apply it when adding members or managing existing members in **Members & Quotas**.
### **Model usage limits**
Team Plan admins can control which AI models each member is allowed to use.
**Model template**
* Go to **Usage → Members & Quotas**
* Find the **Models** column
* Create a model template
![0728 Teammodelmanage]()
1. **Per-member model limits**
For each member, admins can select:
* **All** — all organization-available models
* **Model template** — a reusable model configuration created by the admin
* **Selected models only** — only individually selected models are available
Models outside the configured range are unavailable to that member.
![0418 Teamplan Modellimite Overview]()
2. **Batch member management**
Team Plan admins can manage multiple members at once from **Members & Quotas**.
Admins can:
* Filter members by department based on the organization structure
* Select multiple members or use **the top panel** to set in bulk
* Apply available models or model templates
![0520 Teamplan Batchmodel]()
**Filtering and visibility**
* The **Model Limits** filter helps identify members with restricted model access
* The member table shows the model configuration applied to each member
* Unavailable models appear disabled in project model selectors
**Restriction scope**
* Model restrictions are applied at the member level
* If a restricted member creates a project:
* If the billing mode is **Owner Mode:** everyone in that project can only use the models allowed **under that member’s model configuration**
* If the billing mode is **Presenter Mode**: everyone in that project can use the models allowed **under their own model configuration**
![0418 Teamplan Modellimit Result]()
### **Additional payment when seats are insufficient**
If the required VM seat is unavailable, confirming the member configuration redirects the admin to the payment page.
Payment can be completed through:
* **Budget Balance**, when the organization has sufficient balance
* **Stripe**, when additional payment is required
### **Member on-demand limits**
Organization Owners and Admins can configure an independent on-demand limit for each Team Plan member.
Admins can:
* Set limits while adding members
* Modify limits directly in **Members & Quotas**
* Configure multiple members in bulk
* Apply different limits to individual members
A member’s available AI usage depends on:
* Their configured on-demand limit
* The organization’s remaining Budget Balance
When the member approaches their limit, the Team Plan usage alert rules apply.
### **Modify member plans**
Admins can manage:
* On-demand limits
* Model permissions
* VM configurations
* Team Plan membership
**VM upgrades**
* VM upgrades are supported
* Changes take effect after confirmation
* If an additional seat or payment is required, the admin is directed to the payment flow
* Payment can use Budget Balance when sufficient, or Stripe
**VM downgrades**
* VM downgrades are supported
* Each member can downgrade their VM configuration only **once during an entitlement period**
**Remove member**
* Admins can remove members from Team Plan
* The released seat becomes available for reassignment
* Removing a member does not automatically reduce the number of purchased seats
### **Billing management**
Invoices are available from **Personal Center → Billing**.
![Teamplanbill]()
***
## **Budget Balance**
**Budget Balance** is the organization’s shared balance for Team Plan usage and management.
It replaces the previous **Token Balance** name because the balance can now be used for more than AI token consumption.
Budget Balance can be used for:
* Organization-wide AI usage
* Team Plan renewal
* Purchasing additional seats
* Upgrading member VM configurations
### **Top up Budget Balance**
Organization Owners and Admins can top up Budget Balance from the Usage page.
Available preset amounts include:
* **\$100**
* **\$500**
* **\$1,000**
* **\$2,000**
A custom amount can also be entered.
The top-up panel shows:
* Current balance
* Top-up amount
* Balance after top-up
Budget Balance does not expire.
![0728 Teambugage]()
***
## **Add Seats**
Organization Owners and Admins can purchase additional seats at any time from the Usage page.
Additional seats can use different VM configurations, including:
* **2 vCPU / 4 GB RAM**
* **4 vCPU / 8 GB RAM**
* **8 vCPU / 16 GB RAM**
Payment can be completed through:
* Budget Balance, when sufficient
* Stripe
Additional seats follow the current Team Plan expiration date.
If a seat is added during an active Team Plan period, the seat fee is prorated based on the remaining time in the current period.
***
## **Renewal**
Team Plan can be renewed from the **Usage** page by clicking **Renew**.
During renewal, admins can:
* Select a validity period of **1, 3, 6, or 12 months**
* Increase seats
* Reduce seats
* Configure seat quantities for different VM tiers
* Pay with Budget Balance when sufficient
* Continue through Stripe when needed
Before reducing the number of seats, remove the corresponding members from Team Plan so enough seats are available to remove.
Each Team Plan entitlement period can be renewed **only once** before expiration.
![0728 Teamrenew]()
***
## **Seat and Budget Records**
The Usage page provides detailed records for organization seats and Budget Balance.
### **Seat details**
Admins can view:
* Purchased seats by VM configuration
* Used seats
* Available seats
* Seat changes
Seat records help admins understand current capacity and decide when additional seats are required.
![0728 Teamrecord]()
### **Budget history**
Admins can review Budget Balance transactions and filter them by type, including:
* Top-up
* Team Plan renewal
* Additional seat purchase
* VM upgrade
* VM downgrade
* Refund or balance return
Income and expense records are visually distinguished to make balance changes easier to understand.
![0728 Teamrecord2]()
***
## **Transition Note**
The updated Team Plan changes how existing Team Plan token benefits are managed.
For existing Team Plan customers:
* Used Token in Plan and used Token Limit amounts are combined in the usage display
* Existing configured member Token Limits remain unchanged
* Unused **Token in Plan** balance is transferred to the organization’s **Budget Balance**
* The transferred balance becomes available to all Team Plan members
* The balance can also be used for renewal, additional seats, and VM upgrades
Existing Team Plan users retain the value of their unused benefits through this migration.
![0728 Teammove]()
***
## FAQs
Answers to common questions about Team Plan.
Team Plan can be purchased from **Pricing → Team Plan**.
No. Seats and AI usage balance are purchased separately. Team members consume AI usage from the organization’s shared Budget Balance, subject to their configured on-demand limits.
Budget Balance starts from **\$100**. Preset amounts include \$100, \$500, \$1,000, and \$2,000, and custom amounts are also supported.
No. Budget Balance does not expire.
Budget Balance can be used for AI usage, Team Plan renewal, additional seat purchases, and member VM upgrades.
Team Plan supports **1, 3, 6, and 12 months**.
Yes. Seats are associated with VM configurations, and members can be assigned different VM tiers based on available seats.
Yes. Additional seats can be purchased for 2 vCPU / 4 GB, 4 vCPU / 8 GB, or 8 vCPU / 16 GB configurations.
The added seat follows the current Team Plan expiration date. Its seat fee is prorated based on the remaining time in the active period.
Yes. Admins can filter members by department, select multiple members, and configure on-demand token limits and model access in bulk.
Yes. Admins can separately configure a member’s on-demand token limit, available models or model template, and VM configuration.
CodeBanana redirects the admin to the payment page after confirmation. Payment can use Budget Balance when sufficient, or Stripe when additional payment is required.
Yes. VM downgrades are supported, but each member can downgrade only once during an entitlement period.
Purchased seats are reduced during renewal. Before reducing seats, remove the corresponding members from Team Plan so enough seats are available to remove.
Yes. When the organization has sufficient Budget Balance, it can be used for Team Plan renewal, additional seats, and VM upgrades. Otherwise, payment can continue through Stripe.
Yes. The Add Members and member management flows support filtering members based on the organization’s department structure.
Yes. Admins can create model templates that define available models, model order, and the default model, including Auto mode.
Yes. Admins can apply a model template or manually select the available models for individual members or multiple members.
* If the billing mode is **Owner Mode:** everyone in that project can only use the models allowed **under that member’s model configuration**
* If the billing mode is **Presenter Mode**: everyone in that project can use the models allowed **under their own model configuration**
Team Plan uses separate warning rules:
* A member receives a reminder when their remaining on-demand limit is **10% or less and no more than \$10**
* The organization Owner or Admin receives a reminder when the remaining Budget Balance falls below **10%**
Each user receives the reminder at most once per day. The reminder is evaluated independently for each Team Plan member.
Admins can review seat details and Budget Balance history from the Usage page, including top-ups, renewals, additional seats, and VM changes.
Unused Token in Plan value is transferred to the organization’s Budget Balance. It can then be shared by Team Plan members and used for eligible Team Plan payments.
# Add Agent
Source: https://docs.codebanana.com/documentation/working-in-code-banana/add-agent
Add and share project agents across CodeBanana projects (A2A), including project-named agents, Members-based agent adding, owner-controlled sharing permissions, Can Edit and Can Read access, Personal Agent, @agent cross-project task delivery, Discussion messaging, private chat messaging, contacts lookup, and Team Agent access to Discussion history.
## **A2A — Agent-to-Agent Collaboration**
Add Agent introduces a new **Agent-to-Agent collaboration model** in CodeBanana.
Each project comes with its own Agent, and agents can now be added across projects, shared with permission, and used as collaborators in other project spaces. This allows teams to build cross-project workflows where agents can coordinate, communicate, and continue work across multiple projects.
Each project has its own Agent, and the Agent name is the same as the **project name**.
This makes it easier to identify and reuse agents across projects when:
* Searching for available agents
* Adding an agent through the Members panel
* Mentioning an agent with `@agent`
***
## **Add Agents from the Members Panel**
Agents can be added directly from the **Members** panel in a project.
Available agents may come from two sources:
* **Agents from projects you own**
* **Agents shared with you by other members**, based on the permission scope granted by the agent owner
This allows users to bring both their own agents and authorized shared agents into the current project.
**How to Add an Agent**
In **Chat** or **Project**, click **Members** in the top-right corner.
![0418 Addagent 1]()
In the Members panel, click **Add Agent** to start adding an agent.
![0418 Addagent 2]()
Search for available agents, then select the one you want to add.
After an agent is added, you can mention it with `@agent` to assign tasks or coordinate work across projects.
![0418 Addagent 3]()
***
## **Agent Access Permissions**
The owner of an Agent can define how that Agent may be shared into other projects.
There are three permission levels:
* **Can Edit** Send messages, trigger actions, and modify agent settings after the agent is shared
* **Can Read** View agent output and history after it is shared, but cannot interact with it
* **Disabled** The agent cannot be shared into other projects
These permissions are configured by the **Agent owner** for members in the current group.
***
## **Personal Agent**
Each user has a built-in **Personal Agent** in every organization.
A Personal Agent is the user’s own AI counterpart inside team collaboration. It is not tied to just one temporary project, but follows the user across the projects they participate in within the same organization.
### **Core Characteristics**
* **One Personal Agent per user per organization**
Each user has a separate Personal Agent in each organization. These agents are isolated from one another, and their configuration, memory, and context do not carry across organizations.
* **Automatically joins projects**
When a user is added to a project or Discussion group, their Personal Agent is automatically added as well.
* **Automatically leaves projects**
When a user leaves a project, their Personal Agent leaves the group at the same time.
* **Passive by default**
Personal Agent stays silent unless it is explicitly triggered, such as by being mentioned with `@`.
### **Permission Rules**
Personal Agent is designed as a private user-owned agent space, so interaction permissions depend on who initiates the conversation.
| Initiator | Permission | Description |
| ---------------------------- | ------------ | --------------------------------------------------------------------------------------------------------- |
| **The user themself** | Read + Write | Can trigger both read and write actions, including modifying files inside the Personal Agent project |
| **Other users via @mention** | Read only | Can only retrieve information and receive replies, but cannot trigger file changes or other write actions |
> **Security boundary:** Personal Agent is a private agent space. Only the owner can use it to perform write operations. When other users mention it, it responds in read-only mode.
### **What Personal Agent Is For**
Personal Agent gives each member a persistent AI identity inside the organization.
It can help with:
* Answering on the user’s behalf in project discussions
* Carrying personal preferences and working style across projects
* Organizing personal notes, tasks, and follow-ups
* Querying historical information related to the user’s work
* Supporting personal workflows across multiple projects
### **Typical Use Cases**
* **Answer on the user’s behalf**
Other members can mention a user’s Personal Agent to ask about work that person is responsible for. The agent can respond based on the user’s memory and context.
* **Preserve personal preferences**
Users can teach the agent their working style, collaboration habits, and output preferences, which can then continue to apply across projects.
* **Handle personal workflows**
Users can privately ask the agent to organize notes, review past discussions, plan tasks, or retrieve historical information.
### **Why It Matters**
Personal Agent introduces a personal collaboration layer on top of project collaboration.
Instead of treating AI as something attached only to a project, CodeBanana gives every user an AI counterpart that can participate in teamwork, carry personal context, and respond in a secure and controlled way.
***
## **Agent Collaboration Capabilities**
After being added, an agent can participate more naturally in team workflows.
Supported capabilities include:
* Send messages to **Discussion**
* Send messages directly to private chat contacts
* Query members from **Contacts**
* Read **Discussion** chat history through **Team Agent**
These capabilities help agents better understand project context and collaborate across communication channels.
***
## **Cross-Project Task Delivery**
After an agent is added, you can mention it with `@agent` to assign tasks or coordinate work across projects.
When an Agent receives multiple requests at the same time, requests are handled through the A2A queue mechanism to ensure ordered execution.
This makes it possible to:
* Call another project agent into the current workflow
* Coordinate work between multiple project agents
* Let agents summarize discussions, organize todos, and continue execution across projects
You can see the A2A history by clicking A2A record panel：
![0723 Open A2arecord]()
## **A2A Queue and Stop Mechanism**
When multiple users or agents send requests to the same Agent at the same time, CodeBanana uses a queue mechanism to ensure requests are processed in order.
This applies to:
* Team Agent requests
* A2A requests between projects
Requests are processed based on their arrival time, with no priority override or manual insertion.
### **Queue Rules**
* Requests are processed in the order they are received
* Multiple requests from the same user are also queued sequentially
* The maximum queue size is **50 requests**
* When the queue is full, new requests cannot be added and users will be notified to try again later
* After the current task is completed, the Agent automatically starts processing the next queued request
For queued requests:
* The sender can see the current queue status
* A2A requests receive status updates when:
* The request enters the queue
* The request starts execution
### **Stop Running Tasks**
Users can stop Agent tasks based on their permissions.
| User | Permission |
| -------------------- | ----------------------------------------------------------------- |
| Request sender | Can stop their own requests, including running or queued requests |
| Project / chat owner | Can stop all requests in the current conversation |
| Other users | Cannot stop requests created by others |
After stopping:
* Generated content before stopping is preserved
* The current task is terminated
* The next queued request automatically starts processing
* Users can send new requests, which will enter the end of the queue
### **A2A Stop Behavior**
For cross-project A2A requests, stopping behavior depends on where the request is initiated.
**A2A request from Team Agent**
* Both the source project and target project provide a stop entry
* Stopping from either side only stops that specific response generation
* The two sides do not affect each other
**A2A request from Discussion**
* The stop entry is only available in the target Agent's Team Agent window
* Stopping the task terminates the target Agent response
# Agent modes
Source: https://docs.codebanana.com/documentation/working-in-code-banana/agent-mode
Understand CodeBanana chat modes and agent modes, including Team Agent, Private Ask, Discussion, Discussion Coordinator (DC), Coding mode, read-only agent behavior, shared and individual queue behavior, stopping controls, agent access control, project admins, and team collaboration.
## Project and Chat modes
Each project in CodeBanana has a corresponding group chat that supports three distinct chat modes for different collaboration needs.
### Chat modes
Every project group chat includes three modes:
#### **Team Agent**
Team Agent mode is designed for **shared, controlled collaboration with AI** in a team environment. Instead of each individual interacting with a separate assistant, the team works with a **single, project-level agent** that operates on a shared codebase, shared context, and shared objectives.
This mode ensures that AI-driven changes remain **visible, coordinated, and aligned** across the entire team.
![Chatmodes]()
#### **What is Team Agent**
Team Agent is a **centralized AI collaborator** for the project:
* Operates on the **entire project context** (repo, files, discussions, history)
* Acts as a **shared execution layer** for coding, refactoring, debugging, and task automation
* Ensures all AI-generated changes are **transparent and traceable** within the team workspace
Unlike personal AI tools, Team Agent is not private. Every interaction contributes to the **collective project state**.
#### **Collaboration**
To maintain consistency and avoid conflicts, Team Agent follows a **controlled interaction model**:
* Only one team member can interact with the agent at a time. By default, interaction with the Team Agent is restricted to the **project owner** and **project admins**
* When the agent is idle, project admins can take control without requiring manual approval from the owner
When other members want to use the Team Agent:
* They must **request access (presenter) from the owner**
* Once approved, **only that member is granted interaction rights with the agent**
* During this period, all other members do not have access until control is reassigned
During an active session:
* The agent executes tasks based on the current user’s instructions
* All changes are **immediately visible to the team**
* Usage is attributed to the **project owner**
This model ensures that AI actions are **sequential, reviewable, and non-conflicting**, similar to a “single writer” system in collaborative editing.
![Teamagentrequest]()
#### **Queueing and Stopping**
When multiple users send messages to the **Team Agent**, they are processed through a shared queue.
**Queueing behavior**
* All messages are processed in **send-time order**
* The entire project shares **one Team Agent queue**
* Multiple messages from the same user are placed at the **end of the queue** in sequence
* The queue limit is **50 messages**
* If the queue is full, the user will see: **“Agent is busy, please try again later”**
* After one message finishes, the next queued message starts automatically
* If a queued message is withdrawn, its position is released and later messages move forward automatically
**Queue visibility**
* The Team Agent view shows the current queue status, including how many messages are still waiting
* All members in the project can see the current queue count
**Who can stop a message**
* The **message sender** can stop their own message, whether it is currently running or still waiting in the queue
* The **project owner** can stop any message in the Team Agent queue
* Other members cannot stop messages started by other users
**What happens after stopping**
* Any content already generated is **kept**
* The queue slot is released immediately
* The next queued message starts automatically
* The user can send a new message, which will enter the end of the queue
#### **Agent Modes**
Agent Modes define how the agent interacts with your project, from read-only assistance to direct code execution.
* **Ask-only** Read-only mode. The agent answers questions and provides guidance without modifying any files.
* **Coding** Execution mode. The agent can directly write and edit code within the project.
* **CodeBanana (CB Optimized Model)** Enhanced for project-level tasks with pre-installed capabilities such as deployment.
* **Claude Code (Claude Native Model)** Consistent with the native Claude experience. Requires more explicit prompting and user guidance.
![Agentmode]()
#### **Private Ask**
Private Ask mode (previously My Agent / Private Agent) provides **personal AI assistance within the project**, allowing users to work independently without affecting the shared workflow.
**What is Private Ask**
A private, user-level agent that operates outside of the Team Agent.
It is designed for:
* Asking questions
* Planning tasks
* Exploring ideas
All interactions are **isolated from the team’s active workflow**.
**Usage Characteristics**
* Primarily operates in **ask mode (read-only)**
* Can be used **in parallel with Team Agent sessions**
* Does not modify the shared project
* Usage is attributed according to the project’s billing context
#### **Queueing behavior**
Private Ask uses an **individual queue** instead of the shared Team Agent queue.
* Only your own messages are queued together
* Your Private Ask queue is fully isolated from other users
* Messages are still processed in send-time order
* Stopping a message only affects your own Private Ask session
Private Ask enables users to stay productive individually, while keeping team-level collaboration **structured and conflict-free**.
![Myagent]()
### **Discussion**
Discussion mode is a **team-wide communication space** designed for open collaboration and coordination.
**What is Discussion**
* A shared chat channel where team members can discuss ideas, align on requirements, and coordinate work
* It serves as the **human-first layer of collaboration**, separate from direct execution workflows
* Discussion is also the place where teams can activate a built-in coordination layer when needed
**Usage Characteristics**
* All members can **participate freely**
* Supports **real-time discussions, feedback, and decision-making**
* Conversations are visible to the entire team and remain part of the project context
* By default, Discussion stays focused on communication rather than direct execution
#### **Discussion Coordinator (DC)**
Discussion includes a built-in **Discussion Coordinator (DC)**. DC acts as a coordinator inside group discussion rather than as a direct execution agent.
It can:
* Understand the current discussion context
* Track participants and conversation progress
* Help summarize and organize discussion content
* Route requests to other agents when users mention them in discussion
#### **How DC Works**
DC is **silent by default**. It does not actively interrupt the discussion, and only responds when it is explicitly activated. DC can be activated in two main ways:
* **@mention DC**
* **Reply to a DC message in Discussion**
**Activation paths**
* In the **Team Agent** chat box, mentioning DC activates it through **A2A**, which is useful for multi-agent collaboration
* In the **Discussion** chat box, mentioning DC activates it directly for in-group coordination
* In **Discussion**, replying to a message sent by DC activates it again
**Reply limitations**
* Only replies to messages with the **Discussion** label can reactivate DC
* Replying to messages from other agents does **not** activate DC
Once activated, DC can remain engaged if it determines that the current thread still requires follow-up coordination. When the coordination task is complete, it returns to silent mode automatically.
#### **DC Boundaries**
DC is a coordinator, not an execution agent. It does **not** directly:
* Write code
* Modify files
* Create project content
* Perform write actions in the workspace
Instead, DC helps move discussion forward by:
* Organizing context
* Clarifying requests
* Passing work to the appropriate agent when needed This makes Discussion a lightweight coordination layer, while keeping actual execution in the correct agent mode.
![Disscussion]()
# Basic info
Source: https://docs.codebanana.com/documentation/working-in-code-banana/basicinfo
Configure the agent's Basic Info, including Agent, Soul, Identity, User, and Memory fields, to shape task behavior, communication style, safety rules, personality, user preferences, and project-specific context.
## **What is Basic Info**
Basic Info is a set of structured configurations that shape how the agent:
* Understands tasks
* Makes decisions
* Communicates with users
* Adheres to constraints
Before each response or action, the system injects this information into the agent’s context, making it the **baseline for all reasoning and execution**.
![Basicinfo]()
## **Configuration Fields**
Basic Info consists of several key components:
### Agent
Defines the agent’s working process and execution rules.
For example:
* How to approach tasks
* When to read memory or files
* Coding standards and workflows
```markdown agent.md wrap theme={null}
Workflow Guidelines:
Task intake: Before writing any code, you must first read MEMORY.md and yesterday’s memory/YYYY-MM-DD.md to understand the project background and tech stack.
Code implementation: Prioritize code readability over performance. After each change, you must run a linter.
Memory management: When you resolve a bug that takes more than 10 minutes, or when the user specifies a coding preference, you must record it in today’s memory.
```
### **Soul**
The agent’s core principles and boundaries.
Includes:
* Personality and working style
* Communication tone
* Strict rules (e.g. security constraints, prohibited actions)
```markdown soul.md wrap theme={null}
You are a pragmatic, efficient, and senior full-stack engineer. You avoid over-engineering and follow the principle of “Keep It Simple.”
Communication style: Direct, professional, and concise. No unnecessary explanations.
Hard boundaries:
1. Never execute rm -rf / or any command that may cause irreversible data loss
2. Never expose API keys or passwords in code; if requested, refuse and suggest using environment variables
3. If you are unsure about the safety of code, you must ask the user before proceeding
```
### **Identity**
The agent’s surface-level persona.
Includes:
* Name
* Role definition
* How it introduces itself or interacts with users
```markdown identity.md wrap theme={null}
Name: codebanana
Signature: 🍌
Self-description: Your dedicated, always-on pair programming partner
Greeting style: When starting a new debugging session, you may say:
“Let’s peel this bug apart 🍌”
```
### **User**
Context about the user the agent is serving.
Helps the agent adapt responses based on:
* Technical background
* Preferences
* Communication style
```markdown user.md wrap theme={null}
User profile: A technically proficient product manager building AI agent-based web products
Technical preferences: Familiar with frontend development, frequently uses Next.js, and deploys projects on Vercel or cloud platforms
Communication style: Prefers structured bullet points and fast-paced communication. Avoid explaining basic HTML/CSS concepts—focus directly on architecture and logic
```
### **Memory**
Accumulated knowledge from past interactions.
Maintained by the agent to:
* Store important decisions
* Track preferences or recurring patterns
* Improve future responses
### **Teams**
Defines the team members, roles, and collaboration context that the agent should understand.
Teams helps the agent know:
* Who is involved in the project
* Each member’s role and responsibility
* How different people collaborate
* Which tasks should involve specific team members
This information allows the agent to better coordinate teamwork and provide more relevant assistance in collaborative scenarios.
Example:
```markdown teams.md theme={null}
# Team Members
## Alice
Role: Product Manager
Responsibilities:
- Define product requirements
- Prioritize features
- Coordinate between users and engineering teams
## Bob
Role: Backend Engineer
Responsibilities:
- Maintain backend services
- Design APIs
- Handle database-related tasks
## Design Team
Responsibility:
- Provide UI/UX solutions
- Review product interfaces
```
### **Tools**
Defines local environment information, tool configurations, and project-specific instructions that help the agent use available tools correctly.
Tools provides additional context about:
* Available tools and their intended usage
* Local development environment
* Project-specific commands or configurations
* Important operational notes
This information is included in the agent’s system context and helps prevent incorrect tool usage.
Example:
```markdown tools.md theme={null}
## Environment
- Framework: Next.js
- Package manager: pnpm
- Deployment: Vercel
## Commands
Install dependencies:
pnpm install
Run development server:
pnpm dev
Build project:
pnpm build
## Notes
- Always run tests before creating a pull request
- Do not modify environment variables directly
```
### **Editing and Management**
* Each field can be **individually edited** within the Basic Info panel
* Changes take effect immediately in subsequent agent interactions
* A **reset option** is available to revert to default configurations
![Setbasicinfo 1]()
#### **Why It Matters**
Basic Info ensures that the agent is not just reactive, but **consistent, aligned, and context-aware**:
* Keeps behavior stable across sessions
* Reduces repeated instructions
* Aligns the agent with team workflows and expectations
# Cron job
Source: https://docs.codebanana.com/documentation/working-in-code-banana/cron-job
Automate CodeBanana agents with Cron Jobs, Heartbeat, HEARTBEAT.md, scheduled tasks, daily or weekly runs, custom intervals, file watch triggers, event-based automation, and recurring background workflows.
## **What is Cron Job**
Cron Job is a mechanism to **trigger the agent at the right time**, either:
* At a specific time (e.g. every day at 7 AM)
* At regular intervals (e.g. every 30 minutes)
* When certain events occur (e.g. file changes)
This allows the agent to operate **asynchronously**, even when you are not actively interacting with it.
## **Types of Cron Jobs**
There are three main types:
### **Heartbeat (Interval-based)**
Runs tasks repeatedly at fixed intervals.
* Configure a time interval (e.g. every 30 minutes)
* Executes all tasks defined in **HEARTBEAT.md**
* Suitable for:
* Periodic checks
* Monitoring workflows
* Batch processing
### **Scheduled Tasks (Time/Event-based)**
Runs tasks at specific times or based on triggers.
* **Time-based options**:
* Daily, weekly, monthly
* One-time execution
* Custom intervals
* **Event-based**:
* File Watch（Triggered when files are added, modified, or deleted）
### **Webhook（External Event-based）**
Webhook allows external systems to trigger an agent task through an HTTP POST request.
Instead of running automatically based on time, webhook tasks are executed when an external service sends a request to the generated webhook endpoint.
Common use cases:
* Trigger an agent workflow after a system event
* Connect CodeBanana with external services
* Automate workflows based on third-party events
Examples:
* A CI/CD system triggers an agent to review deployment results
* A monitoring system sends an alert and asks the agent to analyze the issue
* An external application triggers a report generation workflow
### **How to Choose**
1. **Is this a one-time task or reminder?**
* Yes → Use **Scheduled Task**
* No → Continue
2. **Does the task need to run at an exact time?**
* Yes → Use **Scheduled Task**
* No → Continue
3. **Can this task be grouped with other periodic checks?**
* Yes → Use **Heartbeat** (add it to **HEARTBEAT.md**)
* No → Use **Scheduled Task**
4. **Is the task triggered by file changes or events?**
* Yes → Use **Scheduled Task (File Watch)**
* No → Contine
5. **Is the task triggered by external event？**
* Yes → Use **Webhook**
* No → Follow the rules above
#### **Configuration**
**Heartbeat**
1. Enable Heartbeat by turning on the **toggle on the right panel**
2. Set the **execution interval** (e.g. every 10 minutes, 30 minutes, etc.)
3. The system will automatically execute all tasks defined in **HEARTBEAT.md** at each interval
Task management:
* Click **Edit** next to **HEARTBEAT.md** to modify tasks
* Click **Reset** to restore the default configuration
![0723 Cronjobentry]()
![Heartbeat]()
**Scheduled Task**
1. Click **Create New Task** to create a task
2. Define **Trigger Conditions**, which determine when the task runs:
* **Scheduled (time-based)** Choose from:
* Daily — run at a specific time each day
* Weekly — run on a specific day and time
* Monthly — run on a specific date and time
* Once — run a single time
* Every — run at a fixed interval
* **File Watch (event-based)** Triggered when changes occur in a selected directory:
* File created
* File modified
* File deleted Example: trigger a task when an Excel file is deleted from a folder
3. Configure the **Execution Flow**
* Choose a model to execute the task, which can be edited later
* Define what the agent should do when the task is triggered
* This can include any automated action or workflow handled by the agent
![Schaduled]()
**Webhook Task**
1. Select **Webhook** as the trigger type
2. Click **Create Webhook Task**
3. Configure:
* Task Name
* Model
* Execution Flow
![0723 Creatawebhooktask]()
When the endpoint receives an HTTP POST request, the agent automatically starts executing the configured workflow.
![CreateWebhookTask]()
# Dev Tools
Source: https://docs.codebanana.com/documentation/working-in-code-banana/dev-tools
Use CodeBanana Dev Tools from the bottom-left panel, including Terminal commands, Service and port management, VM restart, killing processes, Git Status, commits, commit history, restore, and Deploy for persistent shareable app links.
### **Terminal**
The built-in Terminal allows developers to interact directly with the project environment.
* Execute commands within the project’s runtime environment
* Useful for:
* Installing dependencies
* Running scripts
* Debugging issues
* Designed to support more advanced or custom development workflows
* You can open dev tools in file management column
![Devtool]()
### **Service**
The Service panel manages runtime processes and active ports.
* View all **active services and ports**
* Restart the VM environment if a service fails
* Kill specific processes or ports when needed
This helps ensure your application runs smoothly and can be quickly recovered if issues occur.
![Service]()
### **Git Status**
Git Status provides version control visibility and basic operations.
* View current file changes
* Create commits using the **Save** action
* Access commit history:
* Review previous versions
* Restore to a specific commit if needed
This allows teams to track changes and maintain a clear development history directly within the workspace.
![Gitoverview]()
![Gitnew]()
### **Deploy**
Deploy enables you to publish your project as a **persistent, shareable link**:
* Generates a stable URL that does not expire
* Automatically updates when the project changes
* Best suited for:
* Static websites
* Lightweight applications
It is recommended to verify your project using preview before deploying.
# Channel
Source: https://docs.codebanana.com/documentation/working-in-code-banana/imsetting
Connect CodeBanana project agents to external messaging channels through IM Settings, create and bind bots, and integrate Feishu, Lark, Telegram, Slack, Discord, DingTalk, WeCom, QQ, and Mattermost for agent conversations outside CodeBanana.
## **What is Channel**
Channel is a bridge between CodeBanana and external IM tools.
By configuring a bot, you can send messages from platforms like Slack or Telegram directly to your project agent and receive responses in return.
This enables a more flexible workflow where the agent is accessible **across different communication environments**.
#### **Supported Platforms**
You can connect the agent to multiple IM platforms, including:
* Feishu
* Telegram
* Slack
* Discord
* DingTalk
* WeCom
* QQ
![0723 Channel]()
### **Create a Bot**
To enable Channel, you need to create and configure a bot:
* Click your **avatar → IM Setting → Add Bot**
* Select the target IM platform
* Provide required credentials (e.g. ID and Secret)
* Once configured correctly:
* Status shows as **Connected**
* Invalid configurations will show as **Disconnected**
You can create multiple bots for the same platform if needed.
![Channelexam]()
![Bindabot]()
![0723 Feishudis]()
#### **Bind Bot to Project**
After creating a bot, it must be linked to a specific project:
* Select a bot and **bind it to the project**
* Each project can bind **only one bot**
* Each bot can be bound to **only one project**
Only bots in **Connected** status can be bound.
![Bindbot]()
![0723 Imbindingsecon]()
## **How It Works**
Once connected:
* Messages sent from the external platform are routed to the project Team Agent
* The agent processes the request using the same project context
* Responses are returned to the external platform
This allows teams to interact with the agent **without entering the CodeBanana interface**.
## **IM Integration Guide**
#### **Setup**
Create an application in the developer console:
* Log in to [https://open.feishu.cn/app](https://open.feishu.cn/app)
* Create an enterprise self-built application
* Configure name, description, and avatar
* Record **App ID** and **App Secret**
#### **Permissions**
Add the following:
* `im:message:send_as_bot`
* `im:message:readonly`
#### **Event Subscription**
* Use **long connection mode**
* Subscribe to:` im.message.receive_v1`
#### **Configuration in CodeBanana**
* Go to **Personal Center → IM Settings**
* Enter **App ID** and **App Secret**
#### **Setup**
* Open Telegram and search **@BotFather**
* Send `/start `and `/newbot`
* Set bot name and username (must end with` _bot`)
* Copy the **HTTP API Token**
#### **Configuration in CodeBanana**
* Paste the Token into **IM Settings** e.g. `123456789:ABCdefGhIJKlmNoPQRStuVWxyZ`
#### **Setup**
* Log in to [https://work.weixin.qq.com/wework\_admin/frame#/manageTools](https://work.weixin.qq.com/wework_admin/frame#/manageTools)
* Go to Security and Management → Management Tool
Create Intelligent Zone / Intelligent robot → Create Intelligent robot → Create manually
![Image]()
* Configure name, logo, and visibility （All）→ **Save**
#### **Messaging Configuration**
* Edit the robot → Click **create by API mode**
![Image]()
* Choose keep-alive mode, and get **Bot ID** and **Secret**
![Image]()
* Go back to CodeBanana,
* Go to IM Setting
* Add a new bot
* Configure as follow & **Save**
![Image]()
* **\[Important]** Link a CodeBanana project to use the IM channel
![Image]()
#### **Setup**
* Visit [https://q.qq.com/#/](https://q.qq.com/#/)
* Enter OpenClaw bot creation portal
* Log in via QR code
* Create bot and obtain:
* **App ID**
* **App Secret**
#### **Configuration in CodeBanana**
* Add credentials in **IM Settings**
#### **Setup**
* Open [https://open-dev.dingtalk.com](https://open-dev.dingtalk.com)
* Create OpenClaw bot application
* Record:
* Client ID
* Client Secret
#### **Configuration in CodeBanana**
* Enter credentials in **IM Settings**
#### **Setup**
Create a Slack App from the official console:
* Visit [https://api.slack.com/apps](https://api.slack.com/apps)
* Click **Create New App → From scratch**
* App Name: (e.g. CB)
* Development Slack Workspace: select your workspace
* Click **Create App**
#### **Socket Mode (Required)**
Enable long connection mode:
* Go to **Settings → Socket Mode**
* Turn on **Enable Socket Mode**
* Create a token (e.g. `openclaw-socket`)
* Permission: `connections:write (auto-assigned)`
* Click **Generate** and copy the **App-Level Token** (`xapp-...`)
#### **Bot Permissions**
Configure bot capabilities:
* Go to **Features → OAuth & Permissions**
* Under **Bot Token Scopes → Add an OAuth Scope**, add:
* `app_mentions:read` — read @mentions
* `chat:write` — send messages
* `channels:history` — read public channels
* `groups:history` — read private channels
* `im:history `— read direct messages
* `mpim:history` — read group DMs
* `files:write` — upload files/images
#### **Install & Token**
* Click **Install to Workspace → Allow**
* Copy the **Bot User OAuth Token** (`xoxb-...`)
You should now have two tokens:
* `xapp-...` → App Token (Socket Mode / connection)
* `xoxb-...` → Bot Token (message sending)
#### **App Home Configuration**
* Go to **Features → App Home**
* Under **Show Tabs**, enable:
* **Messages Tab** (allows users to DM the bot)
#### **Event Subscription**
Enable event-based message handling:
* Go to **Features → Event Subscriptions**
* Turn on **Enable Events**
* Under **Subscribe to bot events**, add:
* `app_mention` — triggered when bot is mentioned in channels
* `message.im` — direct messages
* `message.channels `— public channels
* `message.groups` — private channels
* Click **Save Changes**
#### **Configuration in CodeBanana**
* Go to **Personal Center → IM Settings**
* Configure both tokens:
* App Token (`xapp-...`)
* Bot Token (`xoxb-...`)
#### **Create Bot**
Set up a bot via the Developer Portal:
* Visit [https://discord.com/developers/applications](https://discord.com/developers/applications)
* Click **New Application** → name it (e.g. `OpenClaw Bot`) → Create
* Go to **Bot → Add Bot → Yes, do it!**
Retrieve credentials:
* Click **Reset Token** → copy Bot Token (shown once only)
* Example format: `MTk5OTk5OTkyNzY4OTk4NzYxNg...`
#### **Required Intents (Critical)**
Enable privileged intents:
* Scroll to **Privileged Gateway Intents**
* Enable all:
* `Message Content Intent `(read message content)
* `Server Members Intent`
* `Presence Intent`
* Click **Save Changes**
#### **Invite Bot to Server**
Generate an invite link:
* Go to **OAuth2 → URL Generator**
Select scopes:
* `bot`
* `applications.commands `(for slash commands)
Select permissions (minimum required):
* `Read Messages / View Channels`
* `Send Messages`
* `Read Message History`
* `Embed Links`
* `Add Reactions (optional)`
* Copy generated URL → open in browser
* Select your server → **Authorize**
#### **Configuration in CodeBanana**
* Go to **Personal Center → IM Settings**
* Paste the **Bot Token**
# Models
Source: https://docs.codebanana.com/documentation/working-in-code-banana/models
Choose and switch CodeBanana models for Team Agent and Private Ask, including Claude, OpenAI GPT, DeepSeek, Gemini, GLM, Kimi, Doubao, MiniMax, and Qwen, with notes on regional availability and deployment restrictions.
#### **Supported Models**
**Auto mode**
Auto mode automatically selects the most suitable model based on your task requirements.
Instead of manually choosing a model, CodeBanana analyzes your prompt and selects a model that best matches the task, considering:
* Task complexity
* Required capabilities
* Expected output quality
* Model usage cost
![0723 Automode]()
Auto mode helps reduce unnecessary model costs by selecting a more cost-efficient model when possible, while still maintaining appropriate task performance.
It is also recommended for users who are not sure which model to choose. Auto mode provides an easier starting point by automatically matching tasks with suitable models.
**Claude**
* Claude Opus 4.8
* Claude Opus 4.7
* Claude Opus 4.6
* Claude Opus 4.5
* Claude Sonnet 5
* Claude Sonnet 4.6
* Claude Sonnet 4.5
* Claude Sonnet 4
**OpenAI**
* gpt-5.6-sol
* gpt-5.6-terra
* gpt-5.6-luna
* gpt-5.5
* gpt-5.4
* gpt-5.4 mini
**DeepSeek**
* DeepSeek-V4 Pro
* DeepSeek-V4 Flash
**Google Gemini**
* gemini-3.1-pro-preview
* gemini-3.1-flash-lite-preview
**GLM**
* GLM-5.2
* GLM-5.1
**Kimi**
* kimi-k2.7-code
**Doubao**
* doubao-seed-2.1-pro
* doubao-seed-2.0-pro
* doubao-seed-1.8
**MiniMax**
* MiniMax-M3
**Qwen**
* qwen3.7-max
* qwen3.7-plus
* qwen3.6-flash
* qwen3-max
#### **Availability**
* Available models may vary by region due to **privacy policies and regulatory requirements**
* Some models may be restricted or unavailable depending on your deployment environment
# Projects and Chats
Source: https://docs.codebanana.com/documentation/working-in-code-banana/project-and-chat
Work with CodeBanana projects and chats, including project lists, GitHub and local projects, creating projects, project visibility across organizations, external projects, Projects and Chats tabs, User Guide files, adding organization members, inviting external contacts, project admins, member management, Discussion, queue visibility, Forward, Restore, notifications, and Request Support.
## **Projects**
Projects are the core unit of work in CodeBanana, combining code, environment, and collaboration into a single workspace. Each project belongs to a fixed organization, determined when the project is created, and that organization does not change over the project’s lifecycle.
Each project can be accessed from two entry points: **Projects** and **Chats**. These two tabs lead to the same underlying project workspace, but surface different types of activity by default.
#### **Project List**
The project list provides flexible management and navigation:
**Filtering and sorting**:
* By project type, owner, creation time, last activity, etc.
**Project types and structure**:
* Visual indicators for:
* GitHub projects
* Local projects
* Projects from the same repository branch are **grouped under a single parent**
![Projectlist]()
![Projectlist 1]()
**Project notifications Settings**
Each project in the project list supports message-related management actions, allowing users to control visibility and notification behavior.
![0714 Projectmessage]()
**Available actions**
Users can access project actions from the **“…” menu**:
* **Pin / Unpin from top**
* Pin important projects to keep them at the top of the project list
* Unpin restores the default sorting behavior
* **Mark as unread**
* Marks the project as unread
* A red notification badge appears again on the project
* The unread count is included in the total notification count shown in the left navigation bar
* **Mute notifications**
* Mutes notifications from the selected project
* The red notification badge changes to a gray indicator
* Muted projects will no longer contribute unread counts to the left navigation bar
**Notification count**
* New messages generate unread indicators on the corresponding project or chat
* Unread counts are accumulated at multiple levels:
* Individual project/chat level
* Group by Project or Chat level which on left navigation bar
* Muted projects are excluded from notification count calculations
**Project operations**:
* click the "folder" icon at the top-right corner
* Additional actions available via the “…” menu
* Permissions apply to all destructive actions
**File and folder management**:
* Folders: Rename, create, upload, copy path, move, delete, download
* Files: Rename, copy path, move, delete, download
**Relative path copy**:
* Copies the file path relative to the project or workspace root
* Deletion requires **agent-level permissions**
**Constraints**:
* Only the **project owner** can delete a project
* Deletion is allowed **only if the owner has more than one project**
![Operateproject]()
#### **Create Project**
CodeBanana supports multiple ways to initialize a project
**Project sources**:
* Create a local project
* Import from GitHub or GitLab
**Templates**:
* Pre-configured templates based on use cases:
* Include styles, skills, and agent settings
* Option to start from **Custom (empty project)**
![Studio Page]()
#### **Project Visibility Across Organizations**
Project visibility is determined by the relationship between:
* The project’s fixed organization
* The organizations the current member has joined
* The member’s role in that project
**If you are the project owner**
* The project is shown only under its own organization
* Even if you are the owner, the project will not appear when your current working organization does not match the project’s organization
**If you are not the project owner**
The system checks whether the project’s organization overlaps with the organizations you have joined.
**When there is overlap**
* If the project’s organization is included in the organizations you have joined, the project is shown only under that organization
* It will not appear under your other organizations
**When there is no overlap**
* If the project’s organization is not included in any organization you have joined, the project is shown under all organizations you have joined
* In this case, it is marked as **External**
This helps the system preserve organization boundaries whenever possible, while still making outside projects visible when no matching organization is available.
**Dynamic evaluation**
Project visibility is calculated based on your current organization state, rather than stored as a permanent result.
For example:
* A project may initially appear as **External** because none of your joined organizations match the project’s organization
* If you later join that organization, the project will automatically switch to the normal organization-based display and the **External** label will disappear
## **User Guide**
After creating a new project, a **User Guide** directory is automatically generated within the project file tree.
This directory contains detailed documentation for CodeBanana’s core features, including both English and Chinese versions. It covers the main modules in Studio and serves two purposes:
* As a **readable reference** for users to understand product capabilities
* As a **knowledge source** that agents can access during task execution
![Projectuserguide]()
### **Adding Members**
**All members** in a project can expand the project team in two ways, depending on whether collaborators are part of your organization or external.
**Add organization members**
Add **existing members from your organization** directly into the project group chat without notification.
![Inviteorgnew]()
**Invite external contacts**
Bring external collaborators into the project using the **Invite external contacts** feature by email:
* Sends a notification and email invitation to the external collaborator
* They do **not become members of your organization**
* Their access is limited to the project they are invited to
This is ideal for:
* Temporary collaborators
* Clients or stakeholders
* External partners who need visibility or participation without full organizational access
![Inviteexternalnew]()
### **Member Management**
Project membership is managed by the **project owner**, who can access member management by clicking the icon in the top-right corner of the agent panel.
**Remove Members**
The **project owner** can remove members from the project group:
* Removed members will **lose access to the project and its chat**
* This action is typically used to:
* Manage team changes
* Revoke access for external collaborators
**Disband Group**
* The **project owner** can disband the entire project group:
* All members will be **removed from the group chat**
* The collaboration session is effectively **terminated**
* This action is irreversible and should be used with caution
Only the **owner** has permission to perform these actions, ensuring that critical changes to team structure remain **controlled and secure**.
![Groupmanagement]()
#### **Project Admins**
Projects and chats support an **Admin** role to make team collaboration more flexible.
Admins can take control of the **Team Agent** under the same access rules as other authorized users. Previously, members needed the project owner to explicitly grant permission before they could interact with the Team Agent. Now, if a user is a project admin, they can directly take over agent access and start a conversation whenever the agent is not actively responding.
**Admin Permissions**
Project admins can:
* Directly interact with the **Team Agent** when it is idle
* Access the Team Agent under the same control rules as other authorized users
This reduces the need for the owner to manually assign access for every interaction, while still keeping control structured.
**How Admins Are Assigned**
The **project owner** can assign or remove admin privileges from the member management panel.
* Up to **5 admins** can be assigned per project
#### **Support Access**
Projects can request dedicated support:
* Use **Request Support** to bring in technical assistance
* A support member is automatically assigned to the project chat
* During support sessions: All token usage is billed at **1.5× rate**
## **Tabs**
Projects and Chats are two entry points for accessing the same set of project workspaces. They separate agent activity from human discussion, making message signals easier to follow.
### **Projects Tab**
The **Projects** tab emphasizes agent-related activity and opens the selected project in **Team Agent** by default.
You can use this tab when you want to focus on:
* Team Agent conversations
* Agent execution progress
* Project files and agent-driven workflows
Projects with recent **Team Agent** activity are ranked higher in this tab, so items with new agent messages may appear at the top.
### **Chats Tab**
The **Chats** tab emphasizes human discussion activity and opens the selected project in **Discussion** by default.
You can use this tab when you want to focus on:
* Human discussion
* Team coordination
* Discussion-first collaboration
This project is still a full project workspace, with:
* Team Agent
* Private Ask
* Discussion
* File tree and project context
Unlike template-based project creation, this flow is designed for conversation-first collaboration. In this mode, the agent can help:
* Summarize your discussions
* Organize todos
* Generate follow-ups
* Turn conversation into project outputs at any time
Projects with recent **Discussion** messages are ranked higher in this tab, so items with new human messages may appear at the top.
Because the two tabs prioritize different activity types, the same project may appear in a different order in each tab.
![Chat]()
#### **Messaging Features**
Chats support essential messaging capabilities for efficient communication:
* Copy, quote, and forward messages
* Automatic notifications when members join or leave a group
* Search across:
* Organization contacts
* Chat history
**Forwarding behavior**
* Messages can be forwarded to:
* The current group
* Other groups
* Restrictions:
* In **Team Agent**, messages can only be forwarded to destinations with agent access
* In **Private Ask** and **Discussion**, forwarding is unrestricted
* **Restore (agent-related)**
* Available on user prompts in agent conversations
* Allows reverting the conversation, model memory, and project state to a previous point
* Not available in standard chat-only conversations
![Messageaction]()
#### **Queue Visibility**
When multiple messages are sent to the **Team Agent**, the project shows queue status directly in the Team Agent view.
* Team Agent displays how many messages are still waiting to be processed
* All project members can see the current queue count
* This helps the team understand whether the agent is currently busy or still processing earlier messages
Queueing behavior is different in **Private Ask**, where each user has an isolated personal queue.
### **New Messages**
New messages help you quickly identify incoming activity across tabs and notification, so you can stay aligned with your team in real time.
#### **Where New Messages Appear**
New message indicators are shown across key areas:
* **Projects**
* **Chats**
* **Notifications**
These indicators highlight where new interactions or updates have occurred.
#### **Priority Handling**
Not all messages are treated equally:
* **@mentions are prioritized** over general new messages
* Mentioned messages are surfaced first, helping you focus on what requires immediate attention
#### **Ranking Logic**
Projects and Chats may show the same project in a different order.
* In **Chats**, projects with new **Discussion** messages are pushed higher
* In **Projects**, projects with new **Team Agent** messages are pushed higher
This separation helps prevent human discussion from being buried under agent process messages.
![Newmessage]()
## Notification
Notifications keep you informed about important activities across your workspace, including permissions and collaboration events.
Notifications cover key events such as:
* Agent permission requests and approvals
* Organization join requests and status updates
* External project invitation
* Billing and usage limit updates
* Group invitation
All notifications are centralized in the **Notifications panel**, where you can:
* Track actions that require your attention
* Quickly navigate to related projects or contexts
![Notification]()
#### **Notification Visibility**
Notification visibility follows the same organization logic as project visibility.
**Standard project notifications**
* Notifications for regular projects are shown only under the organization that owns the project
* If you switch to another organization, those notifications will not appear there
**External project notifications**
* If a project is treated as **External**, its notifications are shown under every organization you have joined
This keeps notifications aligned with the way projects are surfaced across organizations.
# Project Viewer
Source: https://docs.codebanana.com/documentation/working-in-code-banana/projectViewer
Navigate and edit project files in the CodeBanana Project Viewer, manage tabs, preview Markdown, HTML, JSON, and apps, open live services, add line comments, reference content in chat, inspect diff view, track changes, and share files with external links.
## **File Navigation and Tabs**
When working within a project, opened files are organized into tabs for easy switching:
* Multiple files can be opened simultaneously
* Tabs appear at the top of the viewer
* Right-click on tabs to:
* Close current
* Close others
* Close all
This allows you to manage context efficiently while working across different parts of the project.
![Viewertab]()
## **File View and Preview**
The viewer supports both **code and preview modes**, depending on file type:
**Markdown (md)**
* Switch between **Code** and **Preview**
* Edit directly in preview mode
* Use / to trigger formatting options
* Select a line and **right-click for formatting and editing actions**
![Mdedit]()
![Markdownpreview]()
**HTML / JSON / static files**
* Code view and preview mode available
* Option to open in a new tab
![Htmlpreview]()
**App projects**
* Preview across multiple environments:
* Web
* iOS
* Android
* My Device (via external preview tools)
* Refresh available for real-time updates
## **Project Preview**
When a service is running, the Project Viewer enables **live preview**:
* Automatically detects active ports
* Allows instant access to running applications
* Can also be accessed via the **Service panel**
This creates a “what you see is what you build” experience, similar to real-time document editing—but for applications.
![Projectpreview]()
## **Editing and Interaction**
The Project Viewer supports direct interaction with project content:
* Select content to:
* Add comments
* Reference in chat
* Copy content
* Comments are tied to specific lines and remain visible in context
* A floating comment panel allows quick navigation across all comments
![Comment]()
## **Diff and Change Tracking**
When files are modified, additional views are available:
* **Preview** — current state of the file
* **Diff View** — compare changes against previous versions
This helps teams understand exactly what has changed before committing or sharing updates.
![Diff]()
## **Sharing**
Certain file types (e.g. HTML, Markdown, JSON) support **shareable links**, allowing others to view content externally without accessing the full project.
![Share]()
![Shareexam]()
# Skills
Source: https://docs.codebanana.com/documentation/working-in-code-banana/skills
Create, upload, publish, update, share, and reuse CodeBanana Skills with SKILL.md, scripts, references, assets, Skill Market, public skills, private skills, project skills, Add to Project, organization visibility, and versioned workflows.
## **What is a Skill**
A Skill is a structured module that defines how the agent should perform a specific task.
It typically includes:
* A [**SKILL.md**](http://SKILL.md) file (instructions and behavior definition for the agent)
* Optional resources such as:
* Scripts
* References
* Assets
Skills act as **reusable building blocks**, turning one-time solutions into repeatable capabilities.
#### **Skill Types**
Skills in CodeBanana are organized into three categories:
**Public Skills**
* Available in the Skill Market
* Created and shared by the community
* Can be browsed, added to projects, and shared
**Private Skills**
* Created and managed by individual users
* Only visible to you unless published
* Can be reused across your own projects
**Project Skills**
* Skills that are **currently applied within a specific project**
* Only take effect in that project’s agent context
#### **Using Skills**
You can apply Skills to your project in different ways:
**Add from Skill Market**
* Search for relevant skills in market
* Click **Add to Project** to apply in a project
![0723 Skillentry]()
![0723 Skillinstalling]()
**Create and reuse your own Skills**
* Package your workflow into a Skill by chatting with your agent
* Upload it to your private skill library
* Reuse it across multiple projects by opening a project and adding to this project
**Upload Skills**
* Upload a skill to your **Private** library
* Uploaded skills can be reused across different projects
* This is useful when you want to turn an existing workflow or result into a reusable skill
**Share Skills**
* Generate a shareable link
* Others can import and use the skill in their projects
![Shareskill]()
**Publish Skills**
* Convert private skills into published ones
* Choose the visibility scope when publishing:
* Visible to everyone
* Visible within your organization only
![0723 Manageskillentry]()
![0723 Publishskill]()
![0723 Publishskilldetail]()
**Update Skills**
* Skills support updates after creation
* You can publish a new version when the skill content changes
* A new version number can be added during the update process
#### **Where to Manage Skills**
Skills can be accessed from two entry points:
* **Personal Center → Skills**
* Browse the Skill Market
* Manage private and public skills
* View author information in the skill list
![Skillsentrance]()
* **Project → Agent Config → Skills**
* View and manage skills currently active in the project
* Add or remove skills
![Basicinfo]()
#### **Common Use Cases**
**Find and apply a Skill to a project**
* Go to **Skill Market**
* Search for a relevant skill
* Click **Add to Project** and select the target project
**Upload a Skill to your private library**
* Go to **Agent Config → Skills → Skill Market → Private → Upload**
* Upload your skill to your private library
* The uploaded skill can later be added to other projects
**Create a Skill and reuse it across projects**
* Go to **Agent Config → Skills → Skill Market → Private → Upload**
* Upload your skill to your private library
* In another project:
* Go to **Skills → Skill Market → Private → Add to Project**
**Publish a Skill**
* Go to **Skill Market → Private**
* Select the skill and click **Publish**
* Choose whether the skill should be:
* Visible to everyone
* Visible only within your organization
**Update a Skill**
* Open an existing skill
* Update its content or resources
* Go to skill library and click **this project** tab
* Click **update**
* Publish a new version with an updated **version number**
## **Why Skills Matter**
Skills enable teams to:
* Standardize workflows
* Reduce repeated instructions
* Share best practices across projects
* Turn experience into reusable assets
# TicNote Subscription
Source: https://docs.codebanana.com/documentation/working-in-code-banana/ticnote
Use TicNote Subscription in CodeBanana to connect TicNote Pods, record meetings in Discussion, manage recordings with Team Agent, and automatically receive AI summaries and transcripts in the Recordings workspace.
## **What is TicNote Subscription**
TicNote Subscription is a built-in project that every user gets by default in CodeBanana.
You do not need to create or configure it manually. It is always available from day one and is designed to help you connect TicNote Pods, record meetings, and receive AI-generated summaries and transcripts automatically.
Its core experience combines:
* A built-in **Team Agent**
* A **Discussion** space for device connection and recording
* A **Recordings** workspace for storing summaries and transcripts
***
## **Where to Find It**
You can find **TicNote Subscription** directly in your chat or project list.
Important notes:
* Look for the project named **TicNote Subscription**
* Make sure you are the **Owner** of that project
* Only the TicNote Subscription project that you own can bind to your device and manage your recordings
* You can also invite other members into your TicNote Subscription project for shared collaboration
***
## **What Team Agent Can Do**
TicNote Subscription includes a built-in **Team Agent** that is ready to use immediately.
The Team Agent can:
* Answer questions about TicNote features
* Help you understand hardware and subscription options
* List your recordings
* Open summaries and transcripts
* Rename or delete recording files
* Help manage your TicNote-related content inside the project
### **Typical Questions**
You can ask the Team Agent about:
* TicNote software features, such as AI summary, transcription, translation, podcast generation, and research workflows
* TicNote hardware, including Card, Lite, and Pods
* Subscription plans and pricing
* API Skill installation and integrations
* Connecting TicNote Pods with CodeBanana
### **Recording Management**
The Team Agent can also help you manage files in the **Recordings** folder.
Examples:
* “What recordings do I have?”
* “Show me yesterday’s summary”
* “Open the transcript for the latest meeting”
* “Rename this recording”
* “Delete last week’s meeting recording”
***
## **Recording in Discussion**
The **Discussion** tab acts as the control center for TicNote recording.
Here you can:
* Connect TicNote Pods
* Start and stop recording
* Check sync status
* Receive AI-generated summaries after recording ends
### **Recording Flow**
#### **Connect a device**
When no device is connected, the Discussion view shows a device connection area.
* Click **Connect Device**
* Scan for TicNote Pods
* Connect over BLE / Bluetooth
#### **Connected state and file sync**
After connection:
* The connected device name is displayed
* Online status is shown
* A **Start Recording** button becomes available
* Existing recording files on the device begin syncing automatically
### **Sync behavior**
* You can start recording before all historical files finish syncing
* Recording takes priority over file sync
* File sync pauses automatically while recording is in progress
* Sync resumes automatically after recording upload is complete
#### **Recording state**
While recording:
* The device remains visible in Discussion
* A **Stop Recording** button is shown
* Recording continues in the background
* Historical sync stays paused until the recording finishes
#### **Upload and AI summary**
After recording stops:
* The current recording is uploaded automatically
* Paused sync resumes
* AI generates:
* A meeting summary
* A full transcript
These are then pushed back into **Discussion** as a message card.
***
## **Recording File Structure**
Every recording automatically creates a subfolder inside the **Recordings** directory.
The structure looks like this:
```text theme={null}
TicNote Subscription/
├── .codebanana/
└── Recordings/
├── Product Review 2026-04-17/
│ ├── summary
│ └── transcript
├── Weekly Meeting 2026-04-16/
│ ├── summary
│ └── transcript
└── ...
```
This allows the Team Agent to read and manage your recording files directly.
### **What each file contains**
* **summary** — key decisions, highlights, and action items
* **transcript** — full transcription of the recording
***
## **Built-in TicNote Knowledge**
The Team Agent inside TicNote Subscription includes built-in knowledge about TicNote products and services.
It can answer questions about:
* Software features
* Hardware devices
* Subscription plans
* Integration methods
This makes TicNote Subscription both a recording workspace and a built-in product assistant inside CodeBanana.
# Introduction
Source: https://docs.codebanana.com/index
Learn what CodeBanana is and how it combines collaborative coding, shared cloud environments, AI agents, Git integration, organization management, real-time communication, preview, deploy, and app creation for teams, founders, PMs, designers, teachers, and students.
## **What is CodeBanana?**
**CodeBanana** is an AI-powered collaborative coding platform that lets multiple developers work together live in the same workspace. It combines real-time editing, shared environments, and AI support into one smooth workflow. With CodeBanana, teams can build and iterate together in real time, turning ideas into live applications without switching tools or waiting for deployments. Developers, designers, product managers, and marketers can all collaborate in a single live environment.
**CodeBanana** provides:
* **Integrated Development Environment**: Full-featured code editor with AI assistance
* **Git Integration**: Complete version control with collaborative workflows
* **Virtual Machine Tools**: Cloud-based development environments
* **Organization Management**: Team collaboration and project coordination
* **Real-time Communication**: Instant messaging with AI Agent and team members
* **Analytics & Insights**: Performance tracking and project analytics
### Key features：
**Real-Time Collaborative Coding**
* Watch teammates' code live, edit together instantly, and maintain the full project context, just like Google Docs for code.
* Less waiting, fewer meetings, and faster decisions.
* If any issue comes up, you can pull engineers into the same workspace instantly and fix it together.
**AI as a Teammate**
* Code Banana agent that understands your repository and helps generate, review, and optimize code continuously, even while the team discusses and plans.
* This helps teams solve problems instantly without waiting for senior engineers.
**Shared Development Environments**
* Run and share identical app setups with one click. Eliminate “it works on my machine” issues forever.
* No more “it works on my machine” delays.
**Inclusive Workflow for All Roles**
* Designers, PMs, and QA can view, feedback, and participate in the development process without technical barriers.
**One-Click Deploy, Faster from Idea to Live Product**
* The deployed page is always accessible and ready for use, providing a seamless experience for both creators and viewers.
* Any updates you make are automatically reflected on the deployed page, keeping it always up-to-date.
## Who It’s For
* Tech leads and engineering managers seeking visibility and faster iteration.
* Startup founders and small teams looking to ship faster.
* Product managers who want real-time insight into development progress.
* Coding instructors and students who need a live, collaborative environment for teaching, learning, and real-time code walkthroughs.
## **Quickstart guides**
To create your app on CodeBanana, choose the guide that matches your needs:
Jump-start your project by building on our well made templates.
Starting with an exsiting project from your Github by a single click.
Create a beginning app by using the AI-powered tools within CodeBanana.
## **Workspace features**
CodeBanana provides the following essential app creation tools:
* Team collaboration for building together
* Real-time preview of your app
* Publish in minutes
* AI-assisted app creation
* Version control integration for tracking changes
## Need inspiration?
A Web tying game created by CodeBanana
Real-time e-commerce sales dashboard.
Build your fantastic app now
# User Onboarding
Source: https://docs.codebanana.com/onboarding
Complete CodeBanana onboarding for personal, team, and enterprise users, including creating or joining an organization, signing up and signing in, editing team information, importing department and member structures, previewing organization data, using the built-in CB-FAQ Skill, and accessing the Org Admin Workspace for organization management, usage analysis, and AI maturity insights.
## **Sign Up and Sign In**
Before starting the onboarding process, you'll need to create a CodeBanana account and sign in.
Visit the CodeBanana website and click **Sign Up** in the upper-right corner.
![0715 Signup]()
Fill in the registration form:
* Username
* Email address
* Password
Click **Get Code** to receive a verification code, then enter the code and click **Register**.
![0715 Signupinfo]()
Open the verification email sent by CodeBanana.
* Copy the **4-digit verification code**
* Enter it on the registration page to complete verification
Verification codes expire after **15 minutes**. If you don't receive the email, check your spam folder.
![0715 Emailverify]()
After registration is complete:
* Choose **Log in with Email**
* Enter your email address and password
* Click **Log In**
After signing in successfully, you'll enter the onboarding flow.
![0715 Login]()
## **Personal Onboarding**
New users complete a two-step onboarding flow to personalize their workspace. These selections are directly aligned with project template categories.
Choose how you plan to use CodeBanana:
* **Personal** — individual workflows and tasks
* **Team / Enterprise** — collaborative and organizational usage
![Onboardingchooseperson]()
Select your primary purpose, which maps directly to **project template categories**, ensuring a consistent experience when creating new projects.
![Onboardingchoosepurpose]()
## **Organization Onboarding**
For team and enterprise users, onboarding includes organization setup or joining an existing organization.
### Admin actions
Choose how you plan to use CodeBanana:
* **Create New Organization**
* **Join a Team**
![Onboardingchooseteam]()
1. Enter organization name
* Default is generated from admin email
* Can be modified
2. Configure:
* Roles (Owner / Admin)
* Team size
![Onboardingeditteam]()
Upload organization data through excel templates we provide:
* Department structure
* Member list
![Onboardingteaminfo]()
Before importing, you can use **Structure Preview** to verify:
* Department hierarchy
* Member assignments
![Onboardingpreviewteam]()
After upload:
* Invitation emails are automatically sent
* Members can register and join directly
This step is optional and can be skipped and you can manage structure later in **Contacts** page.
### Member actions
Choose to join an existing organization
![Onboardingjointeam]()
* Search organization by name and choose from search results
* Submit a join request
![Onboardingsendrequest]()
After approval by the organization owner:
* You are automatically added as a member
* Your workspace switches to the selected organization
## **Onboarding Enhancements**
CodeBanana provides several built-in onboarding enhancements to help new users get started more smoothly.
### **CB-FAQ Skill**
Every project comes with a built-in **CB-FAQ** Skill.
This Skill stays in sync with product updates, allowing users to ask the agent direct questions about how to use CodeBanana and get product-related guidance inside the project.
![0418 Onboarding Faq]()
### **Org Admin Workspace**
CodeBanana provides a dedicated **Org Admin Workspace** for organization owners and admins.
This workspace is designed as an intelligent operating console for organization management, usage analysis, and AI maturity insights.
**Entry path**:
* New users may receive this workspace by default during onboarding
* Existing users can create it from **New Project → Template → Data & Analyst**
![0418 Onboarding Orgadmin1]()
![0418 Onboarding Orgadmin2]()
**Permission note**:
* All queries and actions are validated against organization roles and structure
* Only **organization Owners** and **Admins** can access these management capabilities
**Supported capabilities**:
* **Organization structure setup and import/export**
Describe the organization in natural language, let the system structure it, preview the result, and confirm import. Excel import and export are also supported.
* **Department management**
Browse the organization tree, and create, rename, or delete departments. All write actions require confirmation.
* **Member management**
View members by department, assign or update departments, edit member information, and remove members.
* **Invitation management**
Invite new members, and view or cancel pending invitations.
* **AI operations metrics**
Query usage data such as token consumption, active users, and department rankings in natural language.
* **Daily AI maturity recommendations**
Enable automatic daily analysis to generate improvement suggestions and update the dashboard.
* **Manual AI maturity recommendations**
Generate AI maturity suggestions on demand for any selected time range.
# Build Your First App
Source: https://docs.codebanana.com/quickstarts/firstapp
Build your first app in CodeBanana by creating a blank project, describing your idea to the agent, attaching files or screenshots, previewing the generated app, refining with follow-up prompts, fixing issues, and publishing the result.
## **How to use CodeBanana**
Codebanana is your personal app and website builder powered by AI. You do not need any coding or tech skills. Just describe what you want to make, CodeBanana will finish the left.
Specializes in generating new projects and building complex features, which is the flagship feature of CodeBanana.
Agent operates as an **autonomous virtual developer** capable of deeply understanding project context, planning architectural logic, and executing full-stack engineering tasks—transforming natural language requirements directly into production-ready app.
* [Prototype a new business idea](https://610d4b83-888f-44a0-ab34-8f504327cc12.codebanana.app/)
* [Fun games ](https://typing-rain-d44fe89d.codebanana.app/)to pass the time
* A forum for complaining about work
* [Create a personal to-do list](https://todo-calendar-app.codebanana.app/)
* Beautifully crafted PPT
* [Personal blog](https://cce1de98-7364-49aa-ad03-e361fd52059e.codebanana.app/)
Follow the steps below to build your first app with CodeBanana. Refine it until it matches your desired design.
Build your app now
*「Estimated time: 7 minutes」*
![Homepage]( "Homepage")
Navigate to the Home screen above and click **Get started,**
Click the **+** as shown in the following image, then choose **Create Project**:
![Image]( "Image")
Just telling the agent what you want to build in your own words, the agent will take you there.
For example, you can type “a Texas Poker game,” or anything else you have in mind. Share as many details as you like. This first prompt helps get your app started, and you can always add more features or make changes later.
*A prompt is a description of the task you want Agent to perform.*
![Image]( "Image")
You can use **My Agent** model to develop your idea, then copy them to **Team Agent** to build it.
You can include text and file attachments in your prompt. After you have completed the description of all the information, click the Send button or press Enter, and the agent will begin working.
CodeBanana automatically designs your app’s appearance, choosing colors, arranging buttons, and organizing menus to make everything look clean and function smoothly.
![Quickpreview]()
Watch your app appear in the preview window. You can explore your app right away and try out any features by clicking the preview. The preview window will be automatically opened when the task finished.
You can view this page in a new tab; simply click the link shown in the image：
![Quickpreopen]()
![Previewtab]()
Once you identify areas for improvement, you can continue communicating with the AI until your wish is fulfilled,
You can also use screenshots to help describe your ideas; simply tap the attachment button next to the send button, or using Ctrl / command + V.
![Quickattachphoto]()
Build your app now
# Import from Github
Source: https://docs.codebanana.com/quickstarts/github
Import an existing GitHub repository into CodeBanana, authorize GitHub access, choose repositories, clone a project, continue development, run and test the app, and publish it from the CodeBanana workspace.
## **Import your GitHub repository**
*Estimated time: 2 minute*
Some developers may want to import their previous projects into CodeBanana for further iteration and modification. CodeBanana provide a method to import your GitHub repositories and turn them into Apps to run, test, and publish your code.
Click the **+** as shown in the following image, then choose **Import Github Repository：**
![Studio Page]()
Click **Authorize**，
![Github Auth]()
Choose the repository you want to authorize, or, you can authorize all:
![Gihubwhich]()
After authorization, you can choose the repository you want to clone from GitHub.
![Gitclone]()
Take me to CodeBanana to build my app
# Start with templates
Source: https://docs.codebanana.com/quickstarts/templates
Start a CodeBanana project from ready-made templates, choose a template during project creation, customize it with the agent, and quickly turn an example such as a presentation or app template into your own project.
## Create from a template
*Estimated time: 1 minute*
If you are unsure where to start, templates make it easy to get going right away. Just choose a template and customize it to fit your needs.
Follow the steps in this guide to create a project with templates in minutes:
Click the **+** as shown in the following image, then choose **Create Project**:
![Studio Page]()
Choose **a template,**
A template normally includes many configuration you need, like skills, scripts and guides.
![0723 Choosenewtemplate]()
![Create Template]()
By communicating with the agent, the content within the template is transformed into what you need, ultimately becoming your unique app.
Take me to CodeBanana to build my app
# Super Organization
Source: https://docs.codebanana.com/super-org
Explore the CodeBanana super organization philosophy, including why traditional organizations break down as they scale, how agent-first collaboration reshapes structure, and how AI can evolve from an individual tool into a new operating system for teams.
## Super Organization and the CodeBanana Product Philosophy
This page is adapted from Mobvoi CEO’s Super Organization keynote, summarizing the organizational philosophy, system design, and real-world practices behind CodeBanana.
***
## Why Traditional Organizations Break Down
Traditional organizations scale by adding more people.
As headcount grows, management structures become more complex:
* More layers
* More process
* More reporting
* More coordination
This creates a predictable structural problem: the larger the organization becomes, the more information is diluted as it moves across layers.
Typical execution flow often looks like this:
* Requirement
* Planning
* Development
* Testing
* Release
After weeks or months of work, the final result can still feel disconnected from the original intent.
The core issue is not individual execution. It is structural entropy:
* The people who understand the need best are often farthest from execution
* Communication cost rises nonlinearly with team size
* Adding people does not necessarily increase output
* Management can suppress entropy temporarily, but cannot eliminate it structurally
This is the organizational problem CodeBanana is designed to address.
***
## What Is a Super Organization
A **Super Organization** is a new organizational structure built around Agents rather than purely around human coordination.
In this structure:
* Agents reduce information loss between decision and execution
* Communication and execution happen in the same place
* Humans and Agents work in parallel rather than in serial handoff chains
* The organization scales through **people × agents**, not just by adding more people
Compared with traditional organizations, a super organization changes the basic logic of scale:
| Dimension | Traditional Organization | Super Organization |
| ------------------------ | ------------------------ | ------------------------ |
| How it scales | More people | People + Agents |
| Core productive unit | Human | Human × Agent |
| Information flow | Layered decay | Real-time shared context |
| Collaboration model | Serial handoff | Parallel collaboration |
| Entropy trend | Increases with size | Structurally resisted |
| Dependence on management | Very high | Significantly reduced |
This is not a tool upgrade. It is a structural replacement.
***
## Five Characteristics of a Super Organization
A super organization is not defined by automation alone, but by how individual capability and collaboration mechanisms are redesigned together.
### Agent First
Agents are not optional assistants. They are the primary execution layer, while humans remain decision-makers.
### Flatter Organization Structure
Decisions move closer to execution. Fewer layers are required to turn intent into action.
### Full-Stack Individuals
More people are able to handle broader scopes of work, reducing handoff loss between specialized roles.
### Prototype-Driven Workflows
Instead of aligning only through documents or meetings, teams align through working outputs, drafts, and prototypes.
### Human × Agent Collaboration
The whole organization learns to work with AI, not just isolated individuals.
***
## Why CodeBanana Exists
Most existing tools fall into two categories:
| Tool Type | Limitation |
| ------------------------ | ------------------------------------------------ |
| Personal AI coding tools | Strong for individuals, weak for organizations |
| Communication tools | Good for discussion, disconnected from execution |
Personal AI tools can create powerful individuals, but they do not create organizational systems. Communication tools help people talk, but they do not help work execute inside the same context.
CodeBanana is built as the system that connects these two worlds.
Its core idea is simple:
**Communication happens where execution happens.**
This is why CodeBanana is designed not as a single-user tool, but as an operating system for super organizations.
***
## Project as the Basic Unit
In CodeBanana, the basic unit is the **project**.
A project is a three-in-one collaboration structure:
* **Agent** — every project comes with its own Agent
* **Chat** — people collaborate in the same shared context
* **Workspace** — files, runtime, and execution environment live in one place
Any work that needs communication, execution, and accumulation can be turned into a project.
This turns the project into the natural container for organizational work.
***
## The Core Collaboration Model
CodeBanana is designed around a collaboration model where humans and Agents work together continuously.
### Team Agent
Team Agent is the shared project-level Agent.
* Transparent
* Shared
* Controlled
* Able to read and write the project workspace
Only one person controls Team Agent at a time, making execution ordered and reviewable.
### Private Ask
Private Ask is a private, read-only mode for individuals.
* Reads project files, code, and conversation history
* Helps with understanding, planning, and analysis
* Remains invisible to others
Together, Team Agent and Private Ask create two complementary modes:
* Shared execution
* Private reasoning
### Discussion and Agent Tabs
Human communication and Agent execution are separated into two tabs:
| Tab | Purpose |
| ---------- | ------------------------------------------ |
| Discussion | Human discussion, decisions, collaboration |
| Agent | Agent tasks, code, execution results |
This keeps human messages readable and agent process messages structured.
### Two Entry Points, Two Views
The same project can appear differently depending on user intent:
* Enter from **Chats** → Discussion-first
* Enter from **Projects** → Agent-first
The underlying workspace is the same, but the surface experience adapts to the collaboration goal.
***
## Key Capabilities That Support the System
Several product capabilities exist specifically to support the super organization model.
### Forwarding and @ Mentions
Messages, files, people, Agents, and Skills can all be referenced directly in context.
This enables:
* Human discussion to become executable instruction
* Agent results to flow back into group communication
* Cross-project collaboration with preserved context
### Skills
Skills turn individual workflows into reusable organizational assets.
* Preinstalled global skills provide common capability
* Teams can build custom skills
* Skills can be published and reused across projects
This allows one person’s method to become the organization’s method.
### Cron Jobs
Agents can work proactively through scheduled execution, heartbeats, and event-driven triggers.
This turns the agent from a reactive tool into an active collaborator.
### Personal Agent
Every user has a Personal Agent that belongs to the individual rather than to a project.
Personal Agent carries personal context across projects and helps users operate across multiple collaboration spaces.
### A2A: Agent-to-Agent Collaboration
Projects can collaborate through their Agents.
A project can expose its Team Agent to other projects, making the project’s knowledge and execution ability reusable across collaboration spaces.
This means:
* Project-to-project collaboration becomes agent-to-agent collaboration
* Shared organizational intelligence can move across projects
* Visibility and permissions define how open or restricted each project agent should be
### Organization Visibility and Contacts
Agents can understand organizational structure, team roles, and member responsibilities.
This makes it possible for them to:
* Notify the right person
* Route issues to the right owner
* Support role-aware coordination
### Enterprise Controls
For organizations, CodeBanana also provides:
* Organization structure management
* Permission control
* Usage analytics
* Audit and export capabilities
* Team-level governance
These controls make agent collaboration practical in real organizational settings.
***
## Why This Is More Than a Product
CodeBanana is not designed to make one person superhuman.
It is designed to make a team evolve structurally.
That distinction matters.
A “super individual” can achieve much more than before, but still hits a ceiling:
* One person still becomes the bottleneck
* AI remains attached to a single user
* The rest of the organization continues to work the old way
A super organization changes the structure itself:
* Information flows with less loss
* Execution is closer to decision-making
* AI becomes part of the organizational system
* Team capacity grows without relying entirely on management scale
***
## Real-World Practice at Mobvoi
CodeBanana is grounded in internal transformation, not just theory.
Mobvoi’s internal rollout followed a real shift from old organizational patterns to agent-centered collaboration.
### Internal transformation milestones
| Time | Milestone |
| ---------- | -------------------------------------------------------- |
| Early 2025 | Began exploring the super organization model |
| April 2025 | Shared the concept publicly |
| Mid 2025 | Validated the ceiling of the “super individual” approach |
| Late 2025 | Began implementing the new structure |
| Early 2026 | Expanded the system across the company |
### Operational outcomes
| Metric | Result |
| ------------------------ | ----------------------- |
| R\&D productivity | 3× to 4× improvement |
| AI token cost | About 15% of labor cost |
| Full-stack ratio in R\&D | 100% |
| Non-R\&D AI-enabled work | Significantly increased |
These results did not come from replacing humans, but from redesigning the collaboration structure.
***
## Department-Level Practice
The super organization model is not limited to engineering.
### R\&D
* Agents help manage daily reports and risk summaries
* Bugs and feedback can be tracked continuously
* Log analysis and issue routing can be automated through A2A and Cron Jobs
### Sales
* Proposal generation can move from days to minutes
* Multiple Agents can collaborate across projects to produce structured documents quickly
### Marketing
* Non-technical users can build dashboards directly with Agents
* Teams can create internal tools without relying entirely on engineering bandwidth
### HR
* Resume review, interview preparation, and candidate comparison can all be partially agent-assisted
* Hiring workflows become more structured and faster to operate
This shows that the model is organizational, not just technical.
***
## When to Create a Project
A good rule is:
**Anything that needs communication, execution, and accumulation deserves a project.**
Common examples include:
* Department-level ongoing collaboration
* Cross-functional long-term work
* Specialized temporary initiatives
* External client or partner collaboration
* Leadership workstreams
* Knowledge bases and organizational memory
This is how organizational work becomes structured and reusable.
***
## Working Habits That Make the System Work
The product alone is not enough. Teams also need the right working habits.
Useful habits include:
1. Set up project basics clearly, including members, roles, admins, and permissions
2. Upload files and organize folder structure early
3. Start with the Agent before escalating to humans when appropriate
4. Use forwarding and @ mentions to keep information flowing precisely
5. Use Cron Jobs and A2A to automate recurring workflows
6. Choose the right device for the role: mobile for managers, desktop for hands-on work
These habits help the system produce compounding value.
***
## Resistance and Adoption Reality
Adopting this model is not frictionless.
In practice, the biggest requirement is leadership conviction.
Common challenges include:
* Resistance to AI-driven coding or execution
* Discomfort with transparent collaboration
* Reluctance to give demand-side teams more direct operational control
* The need to adapt to a more agent-centered rhythm of work
The lesson from rollout is direct:
People can be filtered, but not always educated into structural change.
Adoption depends on both product capability and organizational willingness.
***
## Boundaries and Applicability
This model is powerful, but it is not universal in every context.
It is generally more suitable for:
* Small and medium-sized teams
* Organizations with at least some technical foundation
* Teams with openness to AI collaboration
* Leaders willing to redesign workflows rather than simply layer tools onto old structures
The point is not to follow a trend blindly, but to find the right rhythm for your organization.
***
## Final Thought
The old structure is collapsing. A new operating system for organizations is emerging.
CodeBanana is built around the belief that AI should not remain trapped in isolated chat windows or single-user tools. It should become part of how teams actually work.
A super organization is not about making one person stronger.
It is about enabling the whole team to evolve.