Quickstart
Orient yourself in the AgentOps skill catalog and choose standalone commands or composed chains like /rpi, /swarm, and /implement.
Overview
quickstart is a journey-wide agent skill that explains the AgentOps slash-command catalog and composition graph—usable whenever a solo builder needs to pick or chain agent workflows before committing to a path.
Install
npx skills add https://github.com/boshu2/agentops --skill quickstartWhat is this skill?
- Composition map from /evolve through /rpi loops into research → plan → pre-mortem → crank → vibe → post-mortem
- /swarm parallelizes research, brainstorm, implement, and council with fresh context per wave
- Standalone primitives documented: /research, /brainstorm, /plan, /implement, /vibe, /pre-mortem, /post-mortem, /council,
- Explicit call-outs for what each skill may trigger next (e.g., /plan may call /pre-mortem, /crank may invoke /swarm)
- Categories group THE MULTIPLIER (/swarm), BUILD (/implement), and VALIDATE (/vibe, /pre-mortem, /post-mortem, /council,
- Documents 10+ standalone primitives plus /evolve, /rpi, /crank, and /swarm in the composition map
- Three catalog groupings: THE MULTIPLIER, BUILD, and VALIDATE
Adoption & trust: 826 installs on skills.sh; 384 GitHub stars; 2/3 security scanners passed (skills.sh audits).
What problem does it solve?
You have dozens of AgentOps commands (/swarm, /rpi, /vibe) but no clear picture of what calls what or which primitive fits your current task.
Who is it for?
Builders adopting boshu2/agentops who want a single composition map before invoking /research, /plan, /implement, or validation skills.
Skip if: Executing a specific audit or code change without following into the named child skills—quickstart only routes you.
When should I use this skill?
User is new to AgentOps, asks what skills exist, how /swarm relates to /implement, or needs the full skill catalog composition map.
What do I get? / Deliverables
You can select a standalone skill or a composed chain (evolve → rpi → swarm → implement) matching your research, build, or validate goal.
- Skill selection guidance from the composition map
- Awareness of optional next commands each primitive may trigger
Recommended Skills
Journey fit
Useful at every journey phase - explore requirements and options before committing to a direction.
Where it fits
Scan BUILD and VALIDATE sections to decide whether to start with /research or /brainstorm for an unknown problem.
Follow the map from /plan to optional /pre-mortem before spawning /crank and /swarm implement waves.
Route finished work to /vibe or /council instead of ad-hoc self-review.
Choose /post-mortem plus knowledge lifecycle after a release regression.
Use /pre-mortem from the VALIDATE bucket before locking a large /plan output.
How it compares
Meta catalog and routing guide—not a workflow skill like /implement or a checker like /vibe.
Common Questions / FAQ
Who is quickstart for?
Indie developers and agent-native teams installing AgentOps who need a readable map of available slash skills and how they chain together.
When should I use quickstart?
At Idea when exploring agent workflows, in Build when choosing /implement vs /swarm, at Ship when picking /vibe or /pre-mortem, and in Operate when running /post-mortem or /release—any time you need the catalog before a deeper skill.
Is quickstart safe to install?
It is documentation-only for routing; still review the Security Audits panel on this page and assess risk from the child skills you invoke afterward.
Workflow Chain
SKILL.md
READMESKILL.md - Quickstart
# Full Skill Catalog Pick what you need. Every skill works standalone. `/swarm` multiplies any of them. ## Composition Map — What Calls What ``` /evolve ──► /rpi (per fitness gap, loops until goals met) │ ▼ /rpi ──► /research → /plan → /pre-mortem → /crank → /vibe → /post-mortem │ ▼ /crank ──► /swarm ──► /implement (×N per wave, fresh context each) │ ▼ /swarm ──► parallelize anything: research, brainstorm, implement, council │ ▼ ┌─────────────────────────────────────────────────┐ │ STANDALONE PRIMITIVES │ │ │ │ /research ─────► may trigger /brainstorm │ │ /brainstorm ───► may spawn /swarm │ │ /plan ─────────► may call /pre-mortem │ │ /implement ────► research + plan + build + vibe│ │ /vibe ─────────► /complexity + /council │ │ /pre-mortem ───► /council (failure simulation) │ │ /post-mortem ──► /council + knowledge lifecycle│ │ /council ──────► parallel judges (multi-model) │ └─────────────────────────────────────────────────┘ ``` ## Skills by Category ``` THE MULTIPLIER VALIDATE /swarm - parallelize /vibe - code quality check anything /pre-mortem - plan validation /post-mortem - full retro + knowledge lifecycle BUILD /council - multi-model judges /implement - single task /release - tag + changelog /crank - multi-issue epic /plan - decompose work KNOWLEDGE /rpi - full lifecycle /knowledge - query learnings /retro - quick-capture EXPLORE /post-mortem - full retro + knowledge /research - deep dive /trace - decision provenance /brainstorm - explore ideas /flywheel - health monitoring /bug-hunt - investigate /complexity - code metrics SESSION /doc - docs, README /handoff - save + resume (--mode=readme), /recover - restore after compaction OSS pack /status - dashboard (--mode=oss) PRODUCT /product - define mission CONTRIBUTE (upstream PRs) /goals - fitness specs /pr-research, /plan, /pr-implement /evolve - goal-driven loop /pr-validate, /pr-prep (PR learnings: /post-mortem --scope=pr) META /quickstart - onboarding CROSS-VENDOR /converter - export to Codex, /codex-team - parallel Codex agents Cursor /openai-docs - OpenAI docs lookup ``` # Getting Started with AgentOps ## What is AgentOps? AgentOps is the operational layer for coding agents. It adds bookkeeping, validation, primitives, and flows to Claude Code so sessions do not restart from zero. Instead of each session starting from scratch, AgentOps captures learnings, tracks work, and feeds validated context into the next cycle. ## What Happens When You Run /quickstart? Quickstart does three things fast: 1. **Pre-flight** — Checks your environment (git, CLI tools, directories) 2. **Product framing** — Explains AgentOps as bookkeeping, validation, primitives, and flows 3. **Next step** — Gives one recommended command based on your current setup ## What the Core Surfaces Do ### Research (`/research`) Discovery primitive. Reads code, understands patterns, maps architecture, and surfaces prior bookkeeping before you change anything. ### Validation (`/validation`, `/council`, `/vibe`) Validation is not one command. `/council` pressure-tests plans and diffs, `/vibe` reviews code quality, and `/validation` closes out finished work and extracts learnings. ### RPI (`/rpi`) Full lifecycle flow. `/rpi` composes discovery, implementation, and validation into one end-to-end run. ## Expected Output at Each Step | Step | What You See | |------|-------------| | Pre-flight | Environment status (git,