Passation de tâche avec MCP : résumer le contexte pour un nouveau responsable
Create comprehensive task handoffs using MCP. This workflow shows you how to pull task descriptions and comments, summarize decisions, list open questions, and draft role-based handoff comments for smooth task transitions.
Ce que ce workflow permet
Task handoff with MCP streamlines ownership transitions:
Résultats clés
- Context gathering: Pull all task information, comments, and history
- Decision summarization: Extract key decisions made during task work
- Question identification: List open questions and unresolved items
- Role-based handoffs: Tailor handoff comments to recipient's role
- Smooth transitions: Ensure new owner has all necessary context
Prérequis
Before using this workflow, ensure you have:
- Corcava MCP server configured in your AI assistant
- API key with read access to tasks and comments
- Task with history (comments, status changes)
- For posting handoff comments: API key with write permissions
Workflow pas à pas
Step 1: Gather Task Context
Start by collecting all task information:
Context Gathering Prompt
What the AI does:
- Calls
get_taskto get task details - Calls
list_task_commentsto get all comments - Analyzes task history and status changes
- Presents complete context
Step 2: Summarize Decisions
Extract key decisions from comments and history:
Decision Summarization Prompt
What the AI does:
- Analyzes comments for decision markers
- Identifies technical and process decisions
- Summarizes decisions with context
- Presents decision summary
Step 3: Identify Open Questions
List unresolved questions and blockers:
Question Identification Prompt
What the AI does:
- Scans comments for unanswered questions
- Identifies blockers and dependencies
- Lists items needing clarification
- Formats as actionable questions
Step 4: Determine Recipient Role
Ask about the recipient to tailor the handoff:
Role Identification Prompt
What the AI does:
- Waits for user to specify recipient role
- Adjusts handoff tone and detail level
- Focuses on relevant context for that role
Step 5: Draft Handoff Comment
Create a comprehensive handoff comment:
Handoff Drafting Prompt
What the AI does:
- Compiles all gathered information
- Formats as structured handoff comment
- Tailors to recipient role
- Presents draft for review
Step 6: Post Handoff Comment (with Confirmation)
⚠️ Always Confirm Before Posting Comments
When posting handoff comments:
- Preview First: Show comment text before posting
- Confirmation Token: Require "CONFIRM" before adding comment
- Verify Task: Confirm which task the comment will be added to
Safe Comment Posting Prompt
What the AI does:
- Shows comment preview
- Waits for "CONFIRM"
- Only after confirmation, calls
add_task_comment - Confirms comment was posted
Complete Workflow Prompts
Here are complete, copy-paste ready prompts:
Full Handoff Workflow
Use this when: You want a complete handoff workflow
Quick Handoff
Use this when: You want a quick handoff without full workflow
Workflow Variants by Role
Variant 1: Engineer to Engineer
Technical handoff with code and implementation details:
What happens: AI creates engineer-focused handoff with technical details
Variant 2: Engineer to Product Manager
Business-focused handoff with status and blockers:
What happens: AI creates PM-focused handoff with business context
Variant 3: Designer to Engineer
Design-to-implementation handoff:
What happens: AI creates design-to-engineering handoff
Example: Complete Handoff Session
Here's a real example of how a handoff session flows:
User Prompt
AI Response (Handoff Draft)
Handoff Comment Draft:
Task Summary:
Implementing user authentication system with email/password registration and login. Current status: in_progress, ~60% complete.
Key Decisions Made:
- Using JWT tokens for authentication (24-hour expiration)
- Password requirements: min 8 chars, uppercase, lowercase, number
- Storing hashed passwords using bcrypt
- Email verification will be added in next phase
Open Questions:
- Should we implement "remember me" functionality?
- What rate limiting should we apply to login attempts?
Next Steps:
- Complete password reset flow
- Add unit tests for auth endpoints
- Set up monitoring for failed login attempts
Ready to post? Type CONFIRM to add this handoff comment to the task.
Safety Patterns
⚠️ Confirmation Rules for Comment Posting
When posting handoff comments:
- Always preview: Show comment text before posting
- Require explicit confirmation: Use "CONFIRM" token before adding comment
- Verify task: Confirm which task the comment will be added to
- Review content: Ensure handoff is accurate and complete before posting
Troubleshooting
Missing Context
Symptom: Handoff lacks important information
Possible causes:
- Task has few or no comments
- Important context is in external documents
Fix: Provide additional context: "Also include: [additional information]"
Too Technical or Too Simple
Symptom: Handoff doesn't match recipient's needs
Possible causes:
- Role wasn't specified clearly
- AI didn't adjust detail level
Fix: Specify role clearly: "The recipient is a [specific role] with [level] experience"
Related Tools
This workflow uses these Corcava MCP tools:
get_task
Get task details and context
list_task_comments
Gather comment history for context
add_task_comment
Post handoff comment
Related Use Cases
Acceptance Criteria
Include acceptance criteria in handoffs
Project Kickoff
Hand off new projects to teams
Streamline Task Handoffs
Connect your AI assistant to Corcava and create comprehensive handoff summaries
