Imported from yunaamelia/vps-rdp-workstation (
.github/AGENTS.md). Install upstream withnpx skills add yunaamelia/vps-rdp-workstation --skill .github. Copyright stays with the author.
COMPONENT: GITHUB CONFIGURATION
Scope: CI/CD workflows, AI agent definitions, prompt templates, coding instructions.
STRUCTURE
.github/
├── workflows/ # 12 GitHub Actions CI/CD pipelines
├── agents/ # 20 AI agent role definitions (.agent.md)
├── instructions/ # 17 coding standard instructions (.instructions.md)
├── prompts/ # 24 reusable prompt templates (.prompt.md)
└── scripts/ # Helper scripts (create_secret_file.py)
WORKFLOWS
| Workflow | Trigger | Purpose |
|---|---|---|
ci.yml |
push/PR | Lint + syntax-check + Molecule tests |
ci-enhanced.yml |
push/PR | Extended CI with additional checks |
ci-parallel.yml |
push/PR | Parallel role testing for speed |
deploy-pipeline.yml |
manual/tag | Staging → production with approval gates |
deploy.yml / deploy-staging.yml |
manual | Direct deployment targets |
security-scan.yml |
weekly cron | Trivy scan, SARIF → GitHub Security tab |
validate-playbooks.yml |
push/PR | Playbook syntax validation |
ai-review.yml |
PR | AI-assisted code review |
weekly-integration.yml |
weekly cron | Full integration test suite |
CONVENTIONS
- Secrets: Dual checking pattern — skip (not fail) when secrets missing for forks/PRs.
- Notifications: Discord webhook on failure. Configure via repository secrets.
- Rollback:
deploy-pipeline.ymltriggers automatic rollback on failure. - Molecule: CI uses
molecule testwith Docker driver; same images as local dev. - Agent files:
.agent.mdsuffix for AI agent definitions;.instructions.mdfor standards.
ANTI-PATTERNS
- Hardcoded secrets: Use GitHub repository secrets, never inline values.
- Missing SARIF upload: Security scan results must go to GitHub Security tab.
