run-task

SkillProductivity

Lets your agent execute one Todo task end to end, writing code and meeting its acceptance criteria.

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Then ask your AI: use the run-task skill

About this skill

Execute a single Todo task through In Progress to Review, meeting every acceptance criterion with tests and vibe-lint checks. Refuses Planning-status tasks. Invoked as /agiflow:run-task <task>. Uses get_task, update_task, create_task_comment.

What this skill tells your AI

The instructions your AI receives, as published by hashgraph-online/awesome-codex-plugins in plugins/AgiFlow/ai-plugin/skills/run-task/SKILL.md and read by ahel’s review.

Invoked as /agiflow:run-task. In hosts without slash-prompts, this skill is triggered by matching intent and drives AgiFlow via its MCP tools.

Usage:

  • /agiflow:run-task <task-slug-or-id> - Execute specific task
  • /agiflow:run-task - List and select from available tasks

Examples:

  • /agiflow:run-task DXX-2 (using slug)
  • /agiflow:run-task 01K8FABMNEJG1XTA9JGHSNFV40 (using ID)
  • /agiflow:run-task (interactive selection)

Guardrails

  • Favor straightforward, minimal implementations first and add complexity only when it is requested or clearly required.
  • Keep changes tightly scoped to the requested outcome within the task scope.
  • A task represents a single, focused unit of work that can be completed in one session.

If a task slug/id is provided, load it with get_task; otherwise list available tasks with list_tasks for selection.


AgiFlow Project Management Guidelines

Follow the shared AgiFlow project-management guidelines in references/agiflow-agents.md — agent assignment, the task status workflow and transitions, work-unit best practices, and the tags strategy apply to this workflow.


Task Status Workflow Tasks move through these statuses in order: Planning → Todo → In Progress → Testing → Review → Done Exception paths: Blocked (requires human intervention), Cancelled (terminal).

IMPORTANT: Planning Status Guard This skill ONLY executes tasks in "Todo" or later status. Tasks in "Planning" have NOT been groomed and are NOT ready for execution. Use backlog-grooming to promote Planning tasks to Todo first.

Steps Track these steps as TODOs and complete them one by one.

1. Task Selection & Loading

If task slug/id NOT provided:

  1. Use list_tasks MCP tool to show available tasks:
    • For the ready queue, filter by status: "Todo" (sorted by priority automatically)
    • Do NOT pick up tasks in "Planning" status — they have not been groomed yet
    • Display: slug, title, priority, assignee, acceptance criteria count
  2. Ask user to select which task to work on.
  3. Once user selects, proceed with the selected task slug/id.

If task slug/id IS provided: 4. Use get_task MCP tool with the provided slug/id to retrieve:

  • Task: title, description, priority, status, acceptance criteria
  • Comments: Previous progress updates and discussions
  1. Review the task details to understand:
    • Complete requirements and deliverables
    • Acceptance criteria that must be met
    • Any blockers or dependencies noted in comments

5b. PLANNING STATUS GUARD (MANDATORY):

  • If task status is "Planning": REFUSE to execute
  • Report: "Cannot execute task [SLUG]: still in Planning status. Use backlog-grooming to promote it to Todo first."
  • STOP execution — do not proceed

2. Start Task Execution

  1. If task status is "Todo", use update_task MCP tool to set status to "In Progress".
  2. Create a TODO list of acceptance criteria in execution order (use TodoWrite tool).
  3. Document execution start in task via update_task devInfo:
    devInfo: {
      currentSession: "<current-session-id>",
      startedAt: "<timestamp>"
    }
    

3. Implement Task

  1. For each acceptance criterion:

    • Review the specific requirement
    • BEFORE editing any code: Use vibe-lint MCP get-file-design-pattern (MANDATORY)
    • Implement the criterion keeping changes minimal and focused
    • AFTER editing code: Use vibe-lint MCP review-code-change (MANDATORY)
    • Update task devInfo with implementation notes:
      devInfo: {
        filesChanged: ["path/to/file.ts:42"],
        testResults: { passed: true, coverage: "85%" },
        notes: "Implementation notes here",
        currentSession: "<session-id>"
      }
      
    • Mark acceptance criterion as checked via update_task
    • Update your TODO list (mark criterion as completed)
  2. Between acceptance criteria:

    • Commit changes with meaningful commit messages
    • Run tests to ensure no regressions
    • Update task devInfo with progress

4. Testing Phase

  1. When ALL acceptance criteria are met:

    • Verify all criteria have status checked
    • Use update_task to move status to "Testing"
    • Run the full test suite for affected code:
      • Unit tests
      • Integration tests
      • Type checking and lint
  2. If tests PASS:

    • Update devInfo with passing test results
    • Proceed to step 13 (Task Completion)
  3. If tests FAIL:

    • Increment retryCount in devInfo:
      devInfo: {
        ...existing,
        retryCount: (existing.retryCount ?? 0) + 1,
        lastFailureReason: "<description of what failed>"
      }
      
    • If retryCount >= maxRetries (default maxRetries = 2):
      • Use update_task to move status to "Blocked"
      • Use create_task_comment to document the repeated failure and what human intervention is needed
      • Stop — do not retry
    • If retryCount < maxRetries:
      • Use update_task to move status back to "Todo"
      • Use create_task_comment to document the failure reason
      • Stop — agent will re-pick from Todo queue on next cycle

5. Task Completion

  1. After tests pass, review all files changed (use git diff).

  2. Draft a PR description for the task (DO NOT create the PR yet - just draft the text):

    • Title format: [TASK-SLUG] Task title (e.g., [DXX-2] Add user authentication)
    • Body should include:
      • Task reference and summary
      • List of changes made
      • Files modified (use file.ts:42 format)
      • Test results
  3. Use update_task to set status to "Review" and save draft PR text and commit message:

    {
      status: "Review",
      devInfo: {
        ...existing,
        draftCommitMessage: "feat(auth): add user authentication\n\n- Add login endpoint with session management\n- Add password validation\n- Add unit tests for auth flow",
        draftPr: {
          title: "[DXX-2] Add user authentication",
          body: "## Summary\n\nImplemented user authentication feature.\n\n## Changes\n\n- Added login endpoint\n- Added session management\n\n## Files Modified\n\n- src/auth/login.ts:42\n- src/auth/session.ts:15\n\n## Test Results\n\nAll tests passing."
        },
        finalNotes: "All acceptance criteria met. Files changed: [...]. Tests passing.",
        completedBy: "<member-id>",
        totalDuration: "45 minutes"
      }
    }
    

    Note on draftCommitMessage: Write a conventional commit message that will be used for the final git commit. Format: type(scope): description with optional body listing changes.

  4. Use create_task_comment to add final summary:

    • Task scope and goals achieved
    • All files created/modified (use file.ts:42 format)
    • Test coverage and results
    • Any follow-up work needed

6. Handle Blockers

  1. If blocked at any point:
    • Use update_task to set status to "Blocked"
    • Use update_task to update devInfo with blocker details:
      devInfo: {
        ...existing,
        blockedReason: "<why the task is blocked>",
        blockedBy: "<memberId or 'system'>"
      }
      
    • Use create_task_comment on task with blocker information explaining what human intervention is needed

Reference

  • Use get_task to review task details before starting
  • Use list_task_comments to see previous progress updates
  • Use update_task to track devInfo and acceptance criteria progress

Integration with Other Tools

  • vibe-lint MCP: ALWAYS use before/after editing files for pattern compliance
  • Scaffolding MCP: Document scaffolded code in task comments

Common Mistakes to Avoid

  • ❌ Starting work without reviewing task details with get_task
  • ❌ Not moving task to "In Progress" before starting
  • ❌ Skipping vibe-lint MCP validation (MANDATORY for every file)
  • ❌ Not tracking devInfo (files changed, tests, session ID)
  • ❌ Marking acceptance criteria checked before actually completing them
  • ❌ Moving to "Review" without going through "Testing" first
  • ❌ Not documenting progress in comments
  • ❌ Forgetting to add file references (file.ts:42 format) in final comment
  • ❌ Not running tests before moving to "Review"
  • ❌ Not committing changes incrementally
  • ❌ Moving to "Blocked" without explaining what human intervention is needed
  • ❌ Skipping retryCount increment when moving failed task back to "Todo"
  • ❌ Starting work on a task in "Planning" status (must be promoted to "Todo" via backlog-grooming first)

Signals

GitHub stars
1k
Forks
316
Last commit
Oct 2026
Advanced
Item type
skill
Key
run-task-hashgraph-online
Source
github.com/hashgraph-online/awesome-codex-plugins