> skills.catalog
AgentOps Skills
Every skill works alone. Flows compose them when you want more structure.
This is the doctrine made operational. Where the 12 factors describe principles, the skills below are the concrete tools that enforce them in your agent runtime.
Core skills
/quickstartFastest setup check and next action
Use it when: First time in a repo, or any time you want a one-line read on what to do next.
/councilFactor: Validate Externally (V)Independent judges review a plan, PR, or decision
Use it when: You want multi-model consensus before committing to a direction.
/researchFactor: Research Before You Build (IV)Codebase context + prior learnings before you change code
Use it when: Starting work in unfamiliar territory, or where past decisions matter.
/pre-mortemFactor: Validate Externally (V)Pressure-test a plan before implementation
Use it when: The change is load-bearing. Run this before /implement.
/implementFactor: Lock Progress Forward (VI)One scoped task, built and validated
Use it when: You know what to build and want it done with gates, not vibes.
/rpiResearch → plan → pre-mortem → build → vibe → post-mortem
Use it when: You want the full discovery-to-validation loop end-to-end.
/vibeFactor: Validate Externally (V)Code-quality and risk review before shipping
Use it when: Before every merge. Also runs automatically at the end of /rpi.
/evolveFactor: Compound Knowledge (VIII)Goal-driven improvement loop with regression gates
Use it when: You have a GOALS.md and want the worst fitness gap fixed automatically.
/dreamFactor: Extract Learnings (VII) · Compound Knowledge (VIII)Overnight knowledge compounding, never mutates source code
Use it when: You want the flywheel to keep turning while you sleep.
Full catalog
Grouped by role. Every entry is a slash command in your agent chat.
Validation
/council/vibe/pre-mortem/post-mortemFlows
/research/plan/implement/crank/swarm/rpi/evolveBookkeeping
/retro/forge/flywheel/compileSession
/handoff/recover/status/trace/provenance/dreamProduct
/product/goals/release/readme/docUtility
/brainstorm/bug-hunt/complexity/scaffold/pushFull reference lives in the AgentOps repo.
The ao CLI
The repo-native control plane behind the skills. Handles retrieval, health checks, compounding, goals, and terminal workflows.
| ao quick-start | Set up AgentOps in a repo |
| ao doctor | Check local health |
| ao demo | See the value path in 5 minutes |
| ao search "query" | Search session history and local knowledge |
| ao lookup --query "topic" | Retrieve curated learnings and findings |
| ao context assemble | Build a task briefing |
| ao rpi phased "fix auth startup" | Run the phased lifecycle from the terminal |
| ao evolve --max-cycles 1 | Run one autonomous improvement cycle |
| ao overnight setup | Prepare private Dream runs |
| ao metrics health | Show flywheel health |
The CLI is optional but recommended. Install instructions on the Install page.
The .agents/ directory
AgentOps writes all bookkeeping to a local .agents/ directory in your repo: learnings, findings, handoffs, pre-mortems, plans, and per-run execution packets. It is plain text — grep it, diff it, review it, commit it, or ignore it. No telemetry, no cloud service, no lock-in.
This is how Factor VII — Extract Learnings and Factor VIII — Compound Knowledge are implemented in practice. The principle is doctrine; the directory is the artifact.