Task Execution Skill

Execute development tasks with full autonomy, from issue selection to completion verification.

When to Use This Skill

User requests autonomous task execution
Need to work through a backlog of GitHub Issues
Implementing features or fixes from structured issues
Running continuous development loops

What This Skill Does

Load Context: Reads project plan from Docs/Plan/ folder
Query Issues: Searches GitHub Issues for ready work
Select Task: Picks highest-priority issue with met dependencies
Implement: Writes code, updates files, runs tests
Verify: Ensures tests pass, no lint errors, changes work
Complete: Closes issue with summary or updates labels
Observe: Creates follow-up issues for discovered work
Loop: Repeats until all issues complete

Step-by-Step Procedure

Phase 1: Context Loading

1. Read Docs/Plan/detailed project description
2. Read Docs/Plan/feature list
3. Load current project state from filesystem
4. Query open GitHub Issues

Phase 2: Issue Selection

1. Filter issues:
   - is:open
   - NOT label:"status: blocked"
   - NOT label:"status: in-progress"
   - Dependencies met (check sub-issues)

2. Sort by priority:
   - priority: critical (first)
   - priority: high
   - priority: medium
   - priority: low

3. Select highest priority issue

Phase 3: Implementation

1. Update issue labels: status: in-progress
2. Read full issue body + details
3. Identify files to modify
4. Implement changes (code, tests, docs)
5. Run tests and linting
6. Verify changes work correctly

Phase 4: Verification Checklist

Before closing issue, verify:
- [ ] Code compiles/runs without errors
- [ ] Tests pass (new tests added if needed)
- [ ] No new lint/type errors
- [ ] Related docs updated
- [ ] Changes committed/staged

Phase 5: Completion

1. Add issue comment with summary:
   - What was done
   - Files changed
   - Tests run + results
   - Follow-up issues created
   - Next steps recommendation

2. Close issue OR update labels to:
   - status: review (if PR needed)
   - status: testing (if QA needed)
   - status: done (if complete)

3. Commit changes with message:
   "Fixes #<issue-number>: <title>"

Phase 6: Observation & Follow-ups

Create GitHub issues for:
- Code smells, duplication, complexity
- Lint/type errors discovered
- Missing test coverage
- Documentation gaps
- Security concerns
- Performance issues

Expected Input

User Request Examples:

"Execute tasks autonomously"
"Work through the backlog"
"@Auto Zen start"
"Continue autonomous execution"

Issue Format (GitHub):

Title: TASK: [Clear action verb + object]

## Description

What needs to be done and why

## Implementation Details

-   Approach
-   Files to modify
-   Edge cases to consider

## Test Strategy

How to verify completion

## Dependencies

Depends on: #123, #124

Labels: type: feature, priority: high, status: approved

Expected Output

Issue Comment:

## Implementation Complete ✅

**What was done:**

-   Implemented feature X in src/feature.ts
-   Added unit tests in tests/feature.test.ts
-   Updated documentation in docs/feature.md

**Files changed:** 5

-   src/feature.ts (+120, -0)
-   tests/feature.test.ts (+50, -0)
-   docs/feature.md (+30, -0)
-   package.json (+1, -1)
-   README.md (+5, -0)

**Tests run:** 24 passed, 0 failed
**Coverage:** 87% (+5%)

**Follow-up issues created:**

-   #456: Refactor feature helper for better testability
-   #457: Add integration tests for feature workflow

**Next steps:**
Ready for code review and merge to main.

New Issues Created:

Follow-up tasks for discovered work
Architecture improvements
Test coverage gaps
Documentation updates

Error Handling

If Tests Fail

1. DO NOT close issue
2. Debug test failures
3. Fix failing tests
4. Re-run tests
5. Only proceed when all tests pass

If Blocker Encountered

1. Add label: status: blocked
2. Document blocker in issue comment
3. Create investigation issue
4. Move to next available issue

If Scope Too Large

1. Comment on issue requesting decomposition
2. Hand off to Zen Planner for breakdown
3. Wait for sub-issues to be created
4. Resume with first sub-issue

Integration with Other Skills

Planning: Uses microtasking for task breakdown
Testing: Uses test-generation for QA
Deployment: Uses cloud-deployment for production

Quality Gates

Before marking task complete:

All tests must pass
Coverage must be ≥80% (or same as before)
No new lint/type errors
Documentation updated
Changes committed

Configuration

Required Plan Files:

Docs/Plan/detailed project description - Project vision
Docs/Plan/feature list - Planned features

Required GitHub Labels:

status: pending, status: approved, status: in-progress
priority: critical, priority: high, priority: medium, priority: low
type: feature, type: bug, type: refactor, etc.

Required Permissions:

Read/write access to repository
Create/update/close GitHub Issues
Create branches and commits
Run CI/CD workflows

References

GitHub Issues API documentation
Project plan in Docs/Plan/
.github/copilot-instructions.md - Full agent instructions

task-execution