Corcava logo Le seul outil métier dont vous avez besoin Corcava
Menu
Solutions
Secteurs
Designers RH & recrutement Cabinets d'avocats Immobilier Conseil Marketing digital Comptabilité E‑commerce Développement logiciel
Produit
Fonctionnalités Intégrations
Blog Télécharger
Ressources
Templates Docs MCP FAQ API Comparaisons Feuille de route Quoi de neuf $ Programme d'affiliation
Tarifs
Essai gratuit de 14 jours

Aucune carte bancaire requise

Toilettage du backlog avec MCP : repérer les tâches obsolètes et les nettoyer

Identify tasks with no recent activity, suggest next actions, and optionally add comments asking for clarification. This workflow helps you keep your backlog clean and actionable by surfacing tasks that need attention or should be archived.

Ce que ce workflow permet

Backlog grooming with MCP helps maintain a healthy task backlog:

Résultats clés

  • Stale task detection: Find tasks with no recent activity or updates
  • Action suggestions: Get AI-recommended next steps for stale tasks
  • Clarification requests: Optionally add comments asking for updates
  • Archive candidates: Identify tasks that should be closed or archived
  • Backlog health: Maintain an actionable, up-to-date task list

Prérequis

Before using this workflow, ensure you have:

Workflow pas à pas

Step 1: Find Tasks with No Recent Activity

Start by identifying tasks that haven't been updated recently:

Stale Task Detection Prompt

"Find all tasks in my workspace that haven't been updated in the last 30 days. For each task, show me: title, project, last updated date, status, and assignee."

What the AI does:

  1. Calls list_tasks to get all tasks
  2. Filters tasks by updated_at date (older than 30 days)
  3. Groups stale tasks by project or status
  4. Presents list of stale tasks with context

Step 2: Check Task Comments for Activity

Verify if tasks have recent comments even if the task itself wasn't updated:

Comment Activity Check Prompt

"For the stale tasks you found, check the comments on each task. Identify which tasks have no comments at all, and which tasks have comments but haven't been updated recently. Show me a summary."

What the AI does:

  1. For each stale task, calls list_task_comments
  2. Checks comment timestamps
  3. Categorizes tasks: no comments, old comments, or truly stale
  4. Presents activity summary

Step 3: Suggest Next Actions

Get AI recommendations for what to do with stale tasks:

Action Suggestion Prompt

"For each stale task, suggest a next action. Options include: close/archive (if obsolete), add comment asking for status update, reassign to someone else, or mark as blocked. Explain why you're suggesting each action."

What the AI does:

  1. Analyzes task details (title, description, status, project)
  2. Considers task age and context
  3. Suggests appropriate actions with reasoning
  4. Presents action plan for review

Step 4: Add Clarification Comments (Optional, with Preview)

⚠️ Always Preview Comments Before Adding

Never let AI add comments without showing what will be posted. Use this pattern:

  • Preview First: Show comment text before executing
  • Confirmation Token: Require "CONFIRM" before calling add_task_comment
  • Review Content: Display comment text for review
  • Verify Task: Confirm task ID and title before commenting

Safe Comment Addition Prompt

"For the stale tasks that need clarification, add a comment asking for a status update. But first, show me exactly what comment you'll add to each task. Wait for me to type CONFIRM before adding any comments. If I don't type CONFIRM, don't add anything."

What the AI does:

  1. Drafts comment text for each task (e.g., "This task hasn't been updated in 30+ days. What's the current status?")
  2. Shows preview: "Task: [Title], Comment: [Text]"
  3. Waits for "CONFIRM"
  4. Only after confirmation, calls add_task_comment for each task
  5. Confirms comments were added

Complete Workflow Prompts

Here are complete, copy-paste ready prompts:

Standard Backlog Grooming

"Help me groom my backlog: 1. Find all tasks that haven't been updated in the last 30 days 2. Check comments on each task to see if there's been any activity 3. Suggest next actions for each stale task (close, comment, reassign, etc.) 4. Show me the list organized by suggested action Don't update anything—just show me the analysis and suggestions."

Use this when: You want to review stale tasks without making changes

Backlog Grooming with Comments

"Groom my backlog and add clarification comments: 1. Find tasks not updated in 30+ days 2. For tasks that need clarification, draft a comment asking for status update 3. Show me the list of tasks and the comments you'll add 4. Wait for CONFIRM before adding any comments Example comment format: 'This task hasn't been updated in [X] days. What's the current status? Should we close this or is it still relevant?'"

Use this when: You want to actively reach out about stale tasks

Archive Candidates

"Find tasks that should probably be archived: 1. Find tasks not updated in 60+ days 2. Check if tasks have any recent comments 3. Identify tasks that are likely obsolete or no longer relevant 4. Suggest which tasks should be closed/archived Don't close anything—just show me the candidates for review."

Use this when: You want to identify tasks to archive

Workflow Variants

Variant 1: Project-Specific Grooming

Focus grooming on a specific project:

"Find stale tasks in the [Project Name] project. Tasks not updated in 30+ days. Suggest next actions for each."

What happens: AI filters by project first, then applies grooming logic

Variant 2: Status-Based Grooming

Focus on tasks in a specific status:

"Find all tasks with status 'open' that haven't been updated in 30+ days. These are likely stale backlog items. Suggest whether to close them, add comments, or update status."

What happens: AI filters by status, then identifies stale items

Variant 3: Assignee-Based Grooming

Find stale tasks assigned to specific people:

"Find tasks assigned to [Person Name] that haven't been updated in 30+ days. Check comments to see if there's been any activity, then suggest next actions."

What happens: AI filters by assignee, checks activity, suggests actions

Suggested Actions for Stale Tasks

Here are common actions the AI might suggest:

✅ Add Clarification Comment

When: Task might still be relevant but needs status update

Comment example: "This task hasn't been updated in 30+ days. What's the current status? Is this still relevant?"

Tool: add_task_comment (with confirmation)

🔄 Reassign Task

When: Original assignee is no longer working on it

Action: Update assignee_id to new owner

Tool: update_task (with confirmation)

📝 Update Status

When: Task status is outdated

Action: Change status to "blocked" or "done" if appropriate

Tool: update_task (with confirmation)

🗑️ Close/Archive

When: Task is obsolete or no longer relevant

Action: Update status to "done" or delete if truly obsolete

Tool: update_task or delete_task (with strong confirmation)

Example: Complete Backlog Grooming Session

Here's a real example of how a backlog grooming session flows:

User Prompt

"Find stale tasks in my backlog. Tasks not updated in 30+ days. Check comments, suggest actions, but don't update anything."

AI Response

Stale Tasks Found (8 tasks)

Tasks Needing Clarification (5 tasks):

  • Implement dark mode (Last updated: 45 days ago, No comments) - Suggested: Add comment asking for status
  • Update user guide (Last updated: 38 days ago, Old comments) - Suggested: Reassign or close

Tasks to Archive (3 tasks):

  • Old feature request (Last updated: 90 days ago, No activity) - Suggested: Close/archive

Next Steps:

Would you like me to draft clarification comments for the 5 tasks that need updates? I'll show you the comments first before adding them.

Safety Patterns

⚠️ Confirmation Rules for Write Operations

When grooming includes adding comments or updating tasks:

  • Always preview: Show comment text or changes before executing
  • Require explicit confirmation: Use "CONFIRM" token before any writes
  • Batch carefully: Show full list of comments/changes before applying
  • One at a time for critical actions: For deletions or major changes, handle individually

En savoir plus sur les workflows d'écriture sécurisés →

Troubleshooting

No Stale Tasks Found

Symptom: AI reports no stale tasks

Possible causes:

  • All tasks have been updated recently
  • Time threshold is too short (try 60 days instead of 30)
  • Tasks are in a different workspace

Fix: Adjust time threshold or check different projects

Too Many Stale Tasks

Symptom: Hundreds of stale tasks found

Possible causes:

  • Backlog hasn't been groomed in a long time
  • Many old tasks that should be archived

Fix: Focus on specific projects or statuses first, then expand

Related Tools

This workflow uses these Corcava MCP tools:

list_tasks

Find tasks by date filters

get_task

Get task details including last updated date

list_task_comments

Check comment activity on stale tasks

add_task_comment

Add clarification comments (with confirmation)

update_task

Update status or reassign (with confirmation)

Related Use Cases

Weekly Planning

Generate weekly plans from tasks

Duplicate Detection

Find and merge duplicate tasks

Start Grooming Your Backlog

Connect your AI assistant to Corcava and keep your backlog clean

Get Started → Tool Reference