· Improve skill collections or a named skill with iterative scoring, lint checks, behavioral tests, peer review. Triggers: 'skill refiner', 'improve skills', 'quality sweep', 'batch improve', 'skill loop', 'target 99%'.
· Build/review AI apps: LLMs, RAG, embeddings, agents, evals, local inference. Triggers: 'llm', 'rag', 'embedding', 'openai sdk', 'agent loop', 'fine-tune', 'ollama', 'vllm'. Not for MCP (use mcp).
· Write/review Ansible playbooks, roles, inventories, Vault, Molecule, AWX/AAP. Triggers: 'ansible', 'playbook', 'role', 'inventory', 'group_vars', 'ansible-lint'.
· Strip AI tells from prose in docs, PRs, emails, and your own replies. Filters every response once loaded; full audit on request. Triggers: 'unslop', 'ai writing', 'sounds like chatgpt', 'llm voice'. Not for code (use anti-slop).
· Audit AI-generated code slop: hallucinated APIs, over-abstraction, duplicate code, test theater, noisy comments. Triggers: 'slop', 'AI-generated code', 'cleanup', 'overengineered'. Not for prose (use anti-ai-prose).
· Administer Arch/CachyOS: pacman, AUR, systemd, boot, desktop, GPU, gaming. Triggers: 'arch linux', 'cachyos', 'pacman', 'paru', 'mkinitcpio', 'hyprland'. Not for Debian/Fedora/NixOS.
· Design/review HTTP APIs for FastAPI, Express, NestJS: REST, OpenAPI, pagination, OAuth/JWT. Triggers: 'fastapi', 'express', 'nestjs', 'openapi', 'pagination', 'idempotency', 'rest api', 'endpoint'. Not for schemas (use databases).
· Browse/scrape web pages with Lightpanda, Playwright MCP, agent-browser, or fetch. Triggers: 'browse', 'scrape', 'headless', 'open url', 'read website', 'fill form', 'crawl'. Not for E2E tests (use testing).