blauwtje/exo
Claude Code plugin with four stages and hooks that refuse risky commands.
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
Copy this rig
# review before running: this installs third-party code $ npx degit blauwtje/exo/agents ./rig-exo/agents $ npx degit blauwtje/exo/skills ./rig-exo/skills $ npx degit blauwtje/exo/hooks ./rig-exo/hooks
MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ npx degit blauwtje/exo/skills/check-docs .claude/skills/check-docs $ npx degit blauwtje/exo/skills/configure .claude/skills/configure $ npx degit blauwtje/exo/skills/design-ui .claude/skills/design-ui $ npx degit blauwtje/exo/skills/edit-skills .claude/skills/edit-skills $ npx degit blauwtje/exo/skills/file-issues .claude/skills/file-issues $ npx degit blauwtje/exo/skills/find-cause .claude/skills/find-cause $ npx degit blauwtje/exo/skills/refactor .claude/skills/refactor $ npx degit blauwtje/exo/skills/remember .claude/skills/remember $ npx degit blauwtje/exo/skills/route-skills .claude/skills/route-skills $ npx degit blauwtje/exo/skills/run-plan .claude/skills/run-plan $ npx degit blauwtje/exo/skills/save-session .claude/skills/save-session $ npx degit blauwtje/exo/skills/ship .claude/skills/ship $ npx degit blauwtje/exo/skills/spec .claude/skills/spec $ npx degit blauwtje/exo/skills/start .claude/skills/start $ npx degit blauwtje/exo/skills/verify .claude/skills/verify $ npx degit blauwtje/exo/skills/write-docs .claude/skills/write-docs
$ curl -fsSL --create-dirs -o .claude/agents/build-task.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/build-task.md $ curl -fsSL --create-dirs -o .claude/agents/build-ui.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/build-ui.md $ curl -fsSL --create-dirs -o .claude/agents/critique-ui.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/critique-ui.md $ curl -fsSL --create-dirs -o .claude/agents/fetch-docs.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/fetch-docs.md $ curl -fsSL --create-dirs -o .claude/agents/fix-review.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/fix-review.md $ curl -fsSL --create-dirs -o .claude/agents/locate-code.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/locate-code.md $ curl -fsSL --create-dirs -o .claude/agents/review-branch.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/review-branch.md $ curl -fsSL --create-dirs -o .claude/agents/run-unit.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/run-unit.md $ curl -fsSL --create-dirs -o .claude/agents/solve-hard.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/solve-hard.md $ curl -fsSL --create-dirs -o .claude/agents/survey-ui.md https://raw.githubusercontent.com/blauwtje/exo/main/agents/survey-ui.md
This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 7,204 rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(rm -rf *)",
"Read(**/*.pem)",
"Bash(rm -rf /)",
"Bash(git push --force:*)",
"Bash(sudo *)",
"Read(.env)",
"Bash(rm -rf /*)",
"Read(./.env.*)",
"Read(~/.aws/**)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(sudo:*)",
"Bash(git reset --hard*)",
"Bash(git reset --hard:*)",
"Read(.env.*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(rm:*)",
"Bash(git rebase *)",
"Bash(wget *)",
"Bash(npm publish:*)",
"Bash(git commit *)",
"Bash(gh pr merge *)"
]
}
} Skills (16)
check-docsconfiguredesign-uiedit-skillsfile-issuesfind-causerefactorrememberroute-skillsrun-plansave-sessionshipspecstartverifywrite-docs
Subagents (10)
| build-task model: sonnet | Builds one plan task. Dispatched by build only. |
| build-ui model: sonnet | Builds one design-ui scope. Dispatched by design-ui only. |
| critique-ui model: opus | Judges one built visual surface. Dispatched by design-ui only. |
| fetch-docs model: sonnet | Researches a pinned external library, API or service, or a document on disk, from first-party sources. Not for this repository's code. |
| fix-review model: sonnet | Repairs one branch review's fix findings. Dispatched by verify only. |
| locate-code model: haiku | Finds where code lives across several files, or after one search failed. Not for a named file or symbol, a fix or a design. |
| review-branch model: sonnet | Reviews one finished plan branch. Dispatched by verify only. |
| run-unit model: sonnet | Builds and lands one plan block. Dispatched by build only. |
| solve-hard model: opus | Carries out one hard prompt. Dispatched by build and find-cause only. |
| survey-ui model: sonnet | Surveys one design-ui redesign. Dispatched by design-ui only. |
Similar rigs
lcsmas/orchestra
Run parallel Claude Code and Codex agents in isolated git worktrees
Pragmatist 2.6k tok ·
poshan0126/dotclaude
The standard .claude/ folder structure for everyday development.
Fort Knox 1.7k tok ·
HumanDealer/vibeproof
Vibe-code anything. Ship like a senior. A drop-in rulebook + skills that turn any AI coding agent (Codex, Claude Code, Cursor) into a disciplined senior engineer — built for domain experts who build with AI.
Pragmatist 2.0k tok ·
JoaoPeixoto72/production-quality-ready
Sistema de qualidade e auditoria com gates declarativos para Antigravity e Claude Code
Skill Collector 2.1k tok ·