Skip to content
Back to skills

Quickstart Boshu2 Agentops 6

ASecurity

Show AgentOps next action.

  • 666 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added September 2, 2026
ai-agentsgobashgitapi

Works with

  • cli
  • api

Security analysis

A92/100
  • mediumUses curl or wget to download content

Pro shows the line behind each finding and how to fix it

Scanned September 2, 2026

npx -y skills add majiayu000/claude-skill-registry --skill quickstart-boshu2-agentops-6 --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Quickstart Boshu2 Agentops 6?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Quickstart Boshu2 Agentops 6
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majiayu000-quickstart-boshu2-agentops-6/badge)](https://www.skillsdirectory.com/skills/majiayu000-quickstart-boshu2-agentops-6)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: quickstart
description: 'Show AgentOps next action.'
---
# $quickstart

> **One job:** Tell a new user what AgentOps does and what to do first. Fast.

**YOU MUST EXECUTE THIS WORKFLOW. Do not just describe it.**

## Execution Steps

### Step 1: Detect setup

```bash
git rev-parse --is-inside-work-tree >/dev/null 2>&1 && echo "GIT=true" || echo "GIT=false"
command -v ao >/dev/null && echo "AO=true" || echo "AO=false"
command -v bd >/dev/null && echo "BD=true" || echo "BD=false"
[ -d .agents ] && echo "AGENTS=true" || echo "AGENTS=false"
[ -d "$HOME/.agents" ] && echo "GLOBAL_AGENTS=true" || echo "GLOBAL_AGENTS=false"
[ -n "${CODEX_THREAD_ID:-}" ] || [ "${CODEX_INTERNAL_ORIGINATOR_OVERRIDE:-}" = "Codex Desktop" ] && echo "CODEX=true" || echo "CODEX=false"
```

### Step 2: Show what AgentOps does

Output exactly this (no additions, no diagrams):

```
AgentOps is the operational layer for coding agents.

It gives your coding agent four things it doesn't have by default:

  Bookkeeping — sessions capture learnings, findings, and reusable context in .agents/
  Validation  — $council, $pre-mortem, and $vibe challenge plans and code before shipping
  Primitives  — skills, execution packets, and the ao CLI give you building blocks for almost any interaction
  Flows       — $research, $implement, $validation, and $rpi can run alone or compose end to end

Key skills: $rpi  $research  $validation  $implement  $council  $pre-mortem  $swarm  $status
Full reference: $quickstart --catalog
```

### Step 3: One next action

Match the first row that applies. Output that message.

| Condition | Message |
|-----------|---------|
| GIT=false + AO=true + GLOBAL_AGENTS=true | "🗂  You're outside a git repo but have a global corpus at `~/.agents`. Global knowledge workflow:\n  1. `$harvest` — scan all `.agents/` across your repos and promote artifacts into `~/.agents/learnings/`\n  2. `$compile` — mine, synthesize, and write an interlinked wiki into `.agents/compiled/` (runs from cwd; set `AGENTOPS_COMPILE_RUNTIME=ollama` or `openai` and the matching API key/host)\n  3. `$knowledge-activation` — turn the compiled corpus into playbooks, a belief book, and runtime briefings for future sessions\n  4. `$status` — flywheel health snapshot\nIf instead you want to start a fresh repo-local project here, `git init` first." |
| GIT=false | "⚠ Not in a git repo. Run `git init` first.\n  (If you meant to work against your global `~/.agents` corpus, run `$quickstart` from a dir with `.agents/` or see `$harvest`, `$compile`, `$knowledge-activation`.)" |
| AO=false + CODEX=true | "📦 Install ao CLI first:\n  brew tap boshu2/agentops https://github.com/boshu2/homebrew-agentops\n  brew install agentops\n  ao quick-start\nThen: `$rpi \"a small goal\"` to run your first cycle.\nUse `$bootstrap` after the core seed when you want PRODUCT.md, README.md, and PROGRAM.md." |
| AO=false | "📦 Install ao CLI first:\n  brew tap boshu2/agentops https://github.com/boshu2/homebrew-agentops\n  brew install agentops\n  ao quick-start\nThen: `$rpi \"a small goal\"` to run your first cycle.\nUse `$bootstrap` after the core seed when you want PRODUCT.md, README.md, and PROGRAM.md." |
| AGENTS=false + CODEX=true | "🌱 ao is installed but not initialized here.\n  Run `ao quick-start` to apply the repeatable core seed. `ao quickstart` is the stable alias.\n  Then run `$bootstrap` only if you want the product/operations layer: PRODUCT.md, README.md, PROGRAM.md/AUTODEV.md, and optional hooks.\nThen: `$rpi \"a small goal\"` to run your first cycle." |
| AGENTS=false | "🌱 ao is installed but not initialized here.\n  Run `ao quick-start` to apply the repeatable core seed. `ao quickstart` is the stable alias.\n  Then run `$bootstrap` only if you want the product/operations layer: PRODUCT.md, README.md, PROGRAM.md/AUTODEV.md, and optional hooks.\nThen: `$rpi \"a small goal\"` to run your first cycle." |
| BD=false + CODEX=true | "✅ Codex plugin path ready.\n  Start with `$rpi \"your goal\"`, `$research <topic>`, or `$status`\n  Default installs are hookless; native hooks are optional with `install-codex.sh --with-hooks`.\n  Legacy explicit fallback commands remain `ao codex ensure-start` once per thread and `ao codex ensure-stop` during closeout when needed.\n  Manual escape hatch: `ao codex status`\n  Want issue tracking? `brew install boshu2/agentops/beads && bd init --prefix <prefix>`" |
| BD=false | "✅ Flywheel active. Start now:\n  `$rpi \"your goal\"` — full $discovery → $crank → $validation pipeline\n  `$validation` — close out recent work and capture learnings\n  `$research <topic>` — explore the codebase\n  Want issue tracking? `brew install boshu2/agentops/beads && bd init --prefix <prefix>`" |
| BD=true + CODEX=true | "✅ Codex full stack ready.\n  `bd ready` — see open work\n  Start with `$rpi \"your goal\"`, `$research <topic>`, or `$status`\n  Default installs are hookless; native hooks are optional with `install-codex.sh --with-hooks`.\n  Legacy explicit fallback commands remain `ao codex ensure-start` once per thread and `ao codex ensure-stop` during closeout when needed.\n  Manual escape hatch: `ao codex status`" |
| BD=true | "✅ Full stack ready.\n  `bd ready` — see open work\n  `$rpi \"your goal\"` — start a new goal from scratch\n  `$status` — see current session state" |

---

Starting a new project? Run `$scaffold <language> <name>` to generate project structure with best practices.

## See Also

- [scaffold](../scaffold/SKILL.md) — Project scaffolding and component generation

## Examples

### First-Time Setup

**User says:** `$quickstart`

**What happens:** Agent detects tools, shows one-line status, gives the single next action to run.

### Already Set Up

**User says:** `$quickstart`

**What happens:** Agent detects full stack is ready and suggests `$rpi "your goal"` or `bd ready`.

## Troubleshooting

| Problem | Solution |
|---------|----------|
| Skills not installed | `bash <(curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install.sh)` |
| Hook activation | AgentOps 3.0 is hookless — there is no `ao` command or flag that installs hooks, and CI is the authoritative gate. Hooks are opt-in and author-it-yourself via the `hooks-authoring` skill. Explicit lifecycle fallback commands remain `ao codex ensure-start` and `ao codex ensure-stop`. |
| Flywheel count is 0 | First session — run `$rpi "a small goal"` to start it |
| Want the full skill catalog | Ask: "show me all the skills" or see `references/full-catalog.md` |

## Reference Documents

- [references/getting-started.md](references/getting-started.md)
- [references/troubleshooting.md](references/troubleshooting.md)
- [references/full-catalog.md](references/full-catalog.md)

## Local Resources

### references/

- [references/full-catalog.md](references/full-catalog.md)
- [references/getting-started.md](references/getting-started.md)
- [references/troubleshooting.md](references/troubleshooting.md)

### scripts/

- `scripts/validate.sh`

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.