- Expanded the system prompt to support a wider range of actions, including general chat and agent management.
- Implemented conversational memory, allowing the assistant to maintain context and respond in natural language.
- Updated the interpret function to accept conversation history, improving command parsing.
- Enhanced the rendering of assistant responses to include command outcomes and maintain conversational flow.
- Updated documentation to reflect new features, including conversational context and debug mode for inspecting raw commands.
* feat: Add Copilot trigger creation support
- Add support for One-Time and Recurring triggers in Copilot
- Extend CopilotAssistantMessageActionPart schema with trigger config types
- Update Copilot instructions with trigger creation examples and guidelines
- Implement trigger action handling in messages.tsx component
- Add trigger icons (⏰ for one-time, 🔄 for recurring) in action cards
- Update workflow reducer to handle trigger creation via existing APIs
- Fix action parser to recognize trigger config types in comment format
- Add async trigger processing using createScheduledJobRule and createRecurringJobRule APIs
Users can now ask Copilot to create triggers with natural language requests like:
'Create a daily report trigger at 9 AM' or 'Set up a one-time reminder for next Friday'
* feat: Enhance Copilot message handling and trigger actions
- Pass projectId to Messages and AssistantMessage components for better context
- Refactor applyAction to handle one-time and recurring triggers with improved error handling
- Update handleApplyAll and handleSingleApply to support async action processing
- Remove deprecated pending trigger logic from workflow editor
This update improves the Copilot's ability to manage triggers and enhances the overall message processing flow.
* refactor: route trigger actions via copilot helper
Keep workflow reducer synchronous by removing trigger jobs from the switch and moving job rule API calls into a dedicated helper in messages.tsx. Cache dynamic imports and guard types so Copilot Apply/Apply All handle trigger creation without touching reducer state.
* feat: Add current time to the copilot context
* added conext of triggers to the copilot along with being able to edit and delete triggers
* bug fix for deleting composio triggers
* Add the edit function that allows editing triggers and lets copilot edit triggers too without losing previous jobs
feat: Add update functionality for recurring and scheduled job rules
- Implemented update actions for recurring job rules and scheduled job rules, allowing users to modify existing rules with new input and scheduling configurations.
- Enhanced the UI components to support editing of job rules, including forms for both creating and updating rules.
- Updated the repository interfaces and MongoDB implementations to handle the new update operations for job rules.
This update improves the flexibility of managing job rules within the application.
* Add trigger context to copilot
feat: Enhance trigger management in Copilot
- Added functionality to search for relevant external triggers using the new `search_relevant_triggers` tool, allowing users to discover available triggers based on toolkit slugs and optional query keywords.
- Updated the Copilot context to include detailed descriptions of various external trigger toolkits, enhancing user guidance for trigger creation.
- Improved the overall trigger handling process, ensuring that users can effectively integrate external triggers into their workflows.
This update significantly enhances the Copilot's capabilities in managing and utilizing external triggers.
* Let copilot add external triggers
feat: Enhance external trigger handling in Copilot
- Added support for flexible schemas in external triggers, allowing configuration changes without stripping any data.
- Introduced a new `onRequestTriggerSetup` callback in the Action component to facilitate trigger setup requests.
- Implemented a modal for trigger configuration, improving user experience when setting up external triggers.
- Updated the ComposioTriggerTypesPanel to auto-select trigger types based on initial configuration.
This update significantly improves the management and setup of external triggers within the Copilot interface.
* External trigger cant be edited so we delete and recreate for this
feat: Improve external trigger handling in Copilot
- Added validation for editing external triggers, ensuring users are informed that existing triggers must be deleted and recreated for changes.
- Updated documentation to clarify the limitations of external trigger modifications.
This update enhances user experience by providing clear guidance on managing external triggers within the Copilot interface.
* preventing message.tsx from ballooning up in size
feat: Refactor Copilot message handling and trigger actions
- Removed deprecated logic for loading scheduled and recurring job actions, streamlining the trigger action process.
- Integrated `useCopilotTriggerActions` hook to manage trigger setup and actions more efficiently.
- Enhanced parsing of action parts to improve handling of triggers and their configurations.
- Updated the UI to reflect changes in action handling, ensuring a smoother user experience.
This update optimizes the Copilot's ability to manage triggers and enhances the overall message processing flow.
* refactor: Simplify trigger filtering in Copilot
- Removed unnecessary filtering logic for triggers based on user queries, streamlining the search process.
- Updated response messages to clarify the context of displayed triggers, enhancing user understanding.
This update improves the efficiency of the trigger search functionality within the Copilot interface.
* Revert "refactor: Simplify trigger filtering in Copilot"
This reverts commit b3d041677c.
* simplify the filtering logic for triggers in copilot
feat: Enhance external trigger creation and search functionality in Copilot
- Introduced a critical flow for adding external triggers, emphasizing minimal user input and UI configuration.
- Updated documentation to clarify the external trigger creation process and provided examples for better guidance.
- Simplified the trigger search logic, ensuring users receive relevant results while maintaining clarity in response messages.
This update improves the user experience by streamlining external trigger management and enhancing the search capabilities within the Copilot interface.
---------
Co-authored-by: tusharmagar <tushmag@gmail.com>
- Composio SDK upgrade (0.1.40 -> 0.1.48) changed COMPOSIO_SEARCH_TOOLS response format
- API now returns main_tools instead of results field
- Updated schema to support both formats for backward compatibility
- Fixes tool search functionality that was broken since August 27th upgrade
Resolves: Tool search returning 'No tools found' despite valid tools being available
* added upload button
* image shows up when attached
* added automatic parsing of the image
* make generate image accept an input image
* move image description to debug message
* disable message sending if the image processing hasnt completed yet
* move to x icon for dismiss
* image description processing is stopped on image dismiss
* minor changes
* Fix prebuilt cards model configuration
- Set all prebuilt card models to blank ("model": "")
- Add model fallback in createProjectFromWorkflowJson to use PROVIDER_DEFAULT_MODEL
- Enables different models for OSS vs managed deployments via environment variable
* Refactor assistant template handling and seeding logic
- Replace upfront seeding of all templates with lazy seeding on demand.
- Introduce a new function to fetch specific templates by ID, applying transformations for prebuilt templates.
- Update the ensureLibraryTemplatesSeeded function to skip updating existing templates and log seeding actions.
- Modify project creation to directly use workflow JSON without parsing for default model application.
- Update prebuilt card model configurations to ensure proper defaults are applied.
- Set all prebuilt card models to blank ("model": "")
- Add model fallback in createProjectFromWorkflowJson to use PROVIDER_DEFAULT_MODEL
- Enables different models for OSS vs managed deployments via environment variable
