task-execution
Execute development tasks autonomously by reading GitHub Issues, implementing solutions, running tests, and managing the full development lifecycle until completion
安装与使用
请根据来源仓库和你的 AI 工具说明安装此技能。先阅读技能文件和依赖要求,再添加到工作流。
查看来源 ↗使用指南技能说明
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
使用示例
查看来源仓库中的使用示例与配置说明。
查看来源 ↗来源历史
- 收录时间
- 2026-01-20
- 索引更新时间
- 2026-01-26
- Branch
- main
收录与同步日期属于本站记录。源码修改历史请查看来源仓库。
GitHub ↗chat Comments (0)
Sign in to join the discussion and leave a comment.
No comments yet. Be the first to share your thoughts!