templates/skills/core/agent-automation/SKILL.md
Mandatory workflow that AI agents MUST execute after EVERY implementation.
npx skillsauth add hivellm/rulebook Agent AutomationInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
CRITICAL: Mandatory workflow that AI agents MUST execute after EVERY implementation.
After completing ANY feature, bug fix, or code change, execute this workflow in order:
Run these checks in order - ALL must pass:
1. Type check (if applicable)
2. Lint (MUST pass with ZERO warnings)
3. Format code
4. Run ALL tests (MUST pass 100%)
5. Verify coverage meets threshold (default 95%)
Language-specific commands: See your language template (TYPESCRIPT, RUST, PYTHON, etc.) for exact commands.
IF ANY CHECK FAILS:
# Check for vulnerabilities (language-specific)
# Check for outdated dependencies (informational)
# Find unused dependencies (optional)
Language-specific commands: See your language template for audit commands.
IF VULNERABILITIES FOUND:
If openspec/ directory exists:
# Mark completed tasks as [DONE]
# Update in-progress tasks
# Add new tasks if discovered
# Update progress percentages
# Document deviations or blockers
# Update ROADMAP.md (if feature is milestone)
# Update CHANGELOG.md (conventional commits format)
# Update feature specs (if implementation differs)
# Update README.md (if public API changed)
ONLY after ALL above steps pass:
⚠️ CRITICAL: All commit messages MUST be in English
git add .
git commit -m "<type>(<scope>): <description>
- Detailed change 1
- Detailed change 2
- Tests: [describe coverage]
- Coverage: X% (threshold: 95%)
Closes #<issue> (if applicable)"
Commit Types: feat, fix, docs, refactor, perf, test, build, ci, chore
Language Requirement: Commit messages must be written in English. Never use Portuguese, Spanish, or any other language.
✅ Implementation Complete
📝 Changes:
- [List main changes]
🧪 Quality Checks:
- ✅ Type check: Passed
- ✅ Linting: Passed (0 warnings)
- ✅ Formatting: Applied
- ✅ Tests: X/X passed (100%)
- ✅ Coverage: X% (threshold: 95%)
🔒 Security:
- ✅ No vulnerabilities
📊 OpenSpec:
- ✅ Tasks updated
- ✅ Progress: X% → Y%
📚 Documentation:
- ✅ CHANGELOG.md updated
- ✅ [other docs updated]
💾 Git:
- ✅ Committed: <commit message>
- ✅ Hash: <commit hash>
📋 Next Steps:
- [ ] Review changes
- [ ] Push to remote (if ready)
Skip steps ONLY when:
Exploratory Code: User says "experimental", "draft", "try"
User Explicitly Requests: User says "skip tests", "no commit"
Emergency Hotfix: Critical production bug
In ALL other cases: Execute complete workflow
If workflow fails 3+ times:
1. Create backup branch
2. Reset to last stable commit
3. Report to user with error details
4. Request guidance or try alternative approach
Complete workflow after EVERY implementation:
Only skip with explicit user permission and document why.
<!-- AGENT_AUTOMATION:END -->research
Author a rulebook task spec interactively — research, draft, ask the user clarifying questions, confirm, then create the tasks in rulebook ready for /rulebook-driver. Use when the user wants to plan/spec a feature before implementing.
development
Behavioral guidelines to reduce common LLM coding mistakes — overcomplication, sloppy refactors, hidden assumptions, weak goals. Use when writing, reviewing, or refactoring code. Auto-applies; invoke explicitly via /karpathy-guidelines or 'follow karpathy discipline'.
data-ai
Autonomous AI agent loop for iterative task implementation (@hivehub/rulebook ralph)
data-ai
Use SQL Server for enterprise relational data storage with advanced features, high availability, and Windows integration.