Corcava logoLe seul outil métier dont vous avez besoinCorcava
Menu
Solutions
Secteurs
DesignersRH & recrutementCabinets d'avocatsImmobilierConseilMarketing digitalComptabilitéE‑commerceDéveloppement logiciel
Produit
FonctionnalitésIntégrations
BlogTélécharger
Ressources
TemplatesDocsMCPFAQAPIComparaisonsFeuille de routeQuoi de neuf$ Programme d'affiliation
Tarifs
Essai gratuit de 14 jours

Aucune carte bancaire requise

Tâches runbook ops avec MCP : convertir une revue d'incident en actions

Transform incident timelines into actionable follow-up tasks using MCP. This ops playbook shows you how to parse an incident review, create follow-up tasks with owners and due dates, and add postmortem-style comments for traceability—with approval steps to ensure accuracy.

Ce que ce workflow permet

Ops runbook workflow with MCP helps you turn incidents into actionable work:

Résultats clés

  • Incident parsing: Extract actionable items from incident timelines
  • Task creation: Create follow-up tasks with clear owners and due dates
  • Postmortem documentation: Add traceable comments linking tasks to incidents
  • Accountability: Assign owners and set deadlines for follow-up work
  • Prevention tracking: Track improvements to prevent future incidents

Prérequis

Before using this workflow, ensure you have:

Workflow pas à pas

Step 1: Parse Incident Timeline

Start by analyzing the incident timeline:

Incident Parsing Prompt

"Parse this incident timeline and identify follow-up actions: [Paste incident timeline] Extract: - Root causes that need addressing - Improvements to prevent recurrence - Documentation updates needed - Monitoring/alerting improvements - Process changes required Show me the list of follow-up actions before creating tasks."

What the AI does:

  1. Analyzes incident timeline
  2. Identifies root causes and gaps
  3. Extracts actionable follow-up items
  4. Presents follow-up actions for review

Step 2: Identify Task Owners and Due Dates

Determine who should own each follow-up task:

Owner Assignment Prompt

"For each follow-up action, determine: - Who should own this task (team/role) - What priority it should have - When it should be due (based on urgency) - Which project/board it belongs to Show me the task plan with owners, priorities, and due dates before creating tasks."

What the AI does:

  1. Analyzes each follow-up action
  2. Determines appropriate owners
  3. Sets priorities and due dates
  4. Presents task creation plan

Step 3: Draft Task Descriptions

Create clear task descriptions with context:

Task Description Prompt

"For each follow-up action, draft a task description: - Title: Clear, actionable task name - Description: What needs to be done and why (link to incident) - Include incident reference in description - Add context about root cause or improvement needed Show me all task descriptions before creating tasks."

What the AI does:

  1. Drafts task titles
  2. Creates detailed descriptions
  3. Links tasks to incident
  4. Presents all tasks for review

Step 4: Create Tasks (After Approval)

Create all follow-up tasks after confirmation:

Create Tasks Prompt (Requires Approval)

"Create all the follow-up tasks: - Create each task in the specified project/board - Assign owners and set due dates - Use the task descriptions we reviewed - Show me confirmation after each task is created I must type 'CONFIRM' or 'YES, CREATE TASKS' before you create any tasks."

What the AI does:

  1. Waits for confirmation token
  2. Calls create_task for each follow-up action
  3. Sets owners, due dates, and descriptions
  4. Confirms each task was created

Step 5: Add Postmortem Comments

Add traceable comments linking tasks to the incident:

Postmortem Comment Prompt (Requires Approval)

"Add a postmortem-style comment to each follow-up task: - Comment: 'Follow-up from incident [ID/date]: [brief incident summary]. Root cause: [cause]. This task addresses: [what it fixes].' - Link back to incident timeline - Include traceability information Show me all comments before posting, then wait for my 'CONFIRM' or 'YES, ADD COMMENTS'."

What the AI does:

  1. Drafts postmortem comments for each task
  2. Links tasks to incident
  3. Shows preview of all comments
  4. Calls add_task_comment after approval

Complete Workflow Prompts

Here are complete, copy-paste ready prompts:

Full Incident-to-Tasks Workflow

"Convert this incident timeline into follow-up tasks: 1. [Paste incident timeline] 2. Parse the timeline and identify follow-up actions (root causes, improvements, documentation, monitoring) 3. For each action, determine owner, priority, due date, and project/board 4. Draft task descriptions linking to the incident 5. Show me all tasks for review 6. Wait for my 'CONFIRM' or 'YES, CREATE TASKS' before creating tasks 7. After tasks are created, add postmortem comments linking each task to the incident (wait for 'CONFIRM' or 'YES, ADD COMMENTS') Important: Always preview tasks and comments before creating/posting, and only proceed after explicit confirmation."

Use this when: You want to convert a complete incident review into tasks

Quick Incident Follow-up

"Create follow-up tasks from this incident: [paste timeline]. Show me the tasks, wait for my 'CONFIRM', then create them."

Use this when: You want a quick task creation from incident

Workflow Variants

Variant 1: Critical Incident Follow-up

For high-severity incidents requiring immediate action:

"Convert this critical incident into high-priority follow-up tasks: - Focus on immediate fixes and prevention - Set urgent due dates (within 1-2 weeks) - Assign to senior team members - Create tasks for: immediate fixes, short-term improvements, long-term prevention Prioritize tasks by urgency and impact."

What happens: AI creates high-priority tasks with urgent deadlines

Variant 2: Postmortem Documentation Tasks

For creating documentation and process improvement tasks:

"Convert this incident into documentation and process improvement tasks: - Create tasks for updating runbooks - Document lessons learned - Update monitoring and alerting - Improve incident response procedures Focus on knowledge capture and process improvement."

What happens: AI creates documentation and process improvement tasks

Variant 3: Technical Debt Follow-up

For incidents revealing technical debt:

"Convert this incident into technical debt tasks: - Identify technical improvements needed - Create refactoring tasks - Plan infrastructure improvements - Address architectural issues Focus on reducing technical debt that contributed to the incident."

What happens: AI creates technical improvement tasks

Example: Complete Incident-to-Tasks Session

Here's a real example of how an incident-to-tasks session flows:

User Prompt

"Convert this incident into follow-up tasks: Incident: Database connection timeout Date: 2026-05-10 Duration: 2 hours Impact: Service unavailable for users Timeline: - 14:00: Database connection pool exhausted - 14:05: Alert triggered, on-call engineer notified - 14:15: Identified connection leak in application code - 14:30: Applied hotfix to restart connection pool - 14:45: Service restored - 15:00: Incident resolved Root causes: - Connection leak in user authentication service - Insufficient connection pool monitoring - No automated connection pool recovery Follow-up needed: - Fix connection leak in auth service - Add connection pool monitoring - Implement automated recovery"

AI Response (Task Plan)

Follow-up Tasks Plan (3 tasks):

Task 1:

  • Title: "Fix connection leak in user authentication service"
  • Description: "Fix connection leak identified during incident on 2026-05-10. Root cause: connections not properly closed in auth service."
  • Owner: Engineering Team
  • Priority: High
  • Due Date: 2026-05-17 (1 week)
  • Project: Infrastructure

Task 2:

  • Title: "Add connection pool monitoring and alerts"
  • Description: "Add monitoring for database connection pool to detect issues early. Follow-up from incident 2026-05-10."
  • Owner: DevOps Team
  • Priority: High
  • Due Date: 2026-05-24 (2 weeks)
  • Project: Infrastructure

Task 3:

  • Title: "Implement automated connection pool recovery"
  • Description: "Implement automated recovery mechanism for connection pool exhaustion. Follow-up from incident 2026-05-10."
  • Owner: Engineering Team
  • Priority: Medium
  • Due Date: 2026-05-31 (3 weeks)
  • Project: Infrastructure

Type 'CONFIRM' or 'YES, CREATE TASKS' to create these tasks.

User Approval

"YES, CREATE TASKS"

AI Action (Tasks Created)

✅ Created task #234: "Fix connection leak in user authentication service"

✅ Created task #235: "Add connection pool monitoring and alerts"

✅ Created task #236: "Implement automated connection pool recovery"

Now adding postmortem comments to link tasks to incident...

AI Postmortem Comments Preview

Postmortem Comments (Ready to Post):

Task #234:

"Follow-up from incident 2026-05-10 (Database connection timeout). Root cause: Connection leak in user authentication service. This task addresses: Fixing the connection leak to prevent pool exhaustion."

Task #235:

"Follow-up from incident 2026-05-10 (Database connection timeout). Root cause: Insufficient connection pool monitoring. This task addresses: Adding monitoring and alerts to detect issues early."

Task #236:

"Follow-up from incident 2026-05-10 (Database connection timeout). Root cause: No automated connection pool recovery. This task addresses: Implementing automated recovery mechanism."

Type 'CONFIRM' or 'YES, ADD COMMENTS' to post these comments.

User Approval

"YES, ADD COMMENTS"

AI Action (Comments Posted)

✅ Posted postmortem comment to task #234

✅ Posted postmortem comment to task #235

✅ Posted postmortem comment to task #236

All follow-up tasks have been created and linked to the incident.

Safety Patterns

✅ Preview and Approval Workflow

This workflow uses preview and approval for all write operations:

  • Parse first: Analyze incident before creating tasks
  • Preview tasks: Show all tasks with owners and due dates before creating
  • Preview comments: Show all postmortem comments before posting
  • Confirmation tokens: Require "CONFIRM" or "YES, CREATE TASKS" / "YES, ADD COMMENTS"
  • Two-step approval: Separate approval for task creation and comment posting

Troubleshooting

No Follow-up Actions Identified

Symptom: AI doesn't identify any follow-up actions

Possible causes:

  • Incident timeline too brief
  • No root causes identified
  • Incident already fully resolved

Fix: Provide more detail: "Identify improvements, documentation needs, and monitoring gaps"

Tasks Not Being Created

Symptom: AI doesn't create tasks even after approval

Possible causes:

  • Confirmation token not recognized
  • API key lacks write permissions
  • Project/board ID incorrect

Fix: Use exact confirmation: "CONFIRM" or "YES, CREATE TASKS". Verify API key has write access and project/board IDs are correct.

Owner Assignment Unclear

Symptom: AI assigns tasks to wrong owners

Possible causes:

  • Team structure not clear from incident
  • Multiple teams involved

Fix: Specify owners: "Assign task to [team/role]" or review and correct owner assignments before creating tasks

Related Tools

This workflow uses these Corcava MCP tools:

create_task

Create follow-up tasks from incidents

add_task_comment

Add postmortem comments for traceability

list_projects

Find project/board for follow-up tasks

list_tasks

Review existing follow-up tasks

Related Use Cases

Risk Review

Use incident follow-ups to identify risks

Meeting Notes to Tasks

Similar workflow for converting notes to tasks

Convert Incidents into Actionable Tasks

Connect your AI assistant to Corcava and turn incident reviews into tracked follow-up work

Get Started →Tool Reference