~ / rigs / toverux / cantrips

toverux/cantrips

🤖 Curated agent skills for AI agents — one coherent engineering loop that compounds: grill → spec → implement → review → commit → learn. Adopt it or fork it!

↗ GitHub ★ 3 MIT updated today collection Claude CodeCodex Claude plugin
stack: TypeScript
share on X
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
CONTEXT TAX · EVERY TURN
~7.7k tokens
Heavy · median rig: 2.3k · breakdown
GUARDRAILS
0/5
No committed guardrails · details
Quality 3/100 Guardrails v2

Copy this rig

# review before running: this installs third-party code
$ npx degit toverux/cantrips/.claude ./rig-cantrips  # inspect, then merge into .claude/

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 toverux/cantrips/.claude/skills/sync-upstream .claude/skills/sync-upstream
$ npx degit toverux/cantrips/skills/afk .claude/skills/afk
$ npx degit toverux/cantrips/skills/codebase-design .claude/skills/codebase-design
$ npx degit toverux/cantrips/skills/commit .claude/skills/commit
$ npx degit toverux/cantrips/skills/compound .claude/skills/compound
$ npx degit toverux/cantrips/skills/deputy .claude/skills/deputy
$ npx degit toverux/cantrips/skills/diagnosing-bugs .claude/skills/diagnosing-bugs
$ npx degit toverux/cantrips/skills/grilling .claude/skills/grilling
$ npx degit toverux/cantrips/skills/handoff .claude/skills/handoff
$ npx degit toverux/cantrips/skills/implement .claude/skills/implement
$ npx degit toverux/cantrips/skills/improve-codebase-architecture .claude/skills/improve-codebase-architecture
$ npx degit toverux/cantrips/skills/prototype .claude/skills/prototype
$ npx degit toverux/cantrips/skills/questionnaire .claude/skills/questionnaire
$ npx degit toverux/cantrips/skills/research .claude/skills/research
$ npx degit toverux/cantrips/skills/review-gate .claude/skills/review-gate
$ npx degit toverux/cantrips/skills/roadmap .claude/skills/roadmap
$ npx degit toverux/cantrips/skills/setup-cantrips-loop .claude/skills/setup-cantrips-loop
$ npx degit toverux/cantrips/skills/setup-git-guardrails .claude/skills/setup-git-guardrails
$ npx degit toverux/cantrips/skills/simplify .claude/skills/simplify
$ npx degit toverux/cantrips/skills/spec .claude/skills/spec
$ npx degit toverux/cantrips/skills/storage .claude/skills/storage
$ npx degit toverux/cantrips/skills/tdd .claude/skills/tdd
$ npx degit toverux/cantrips/skills/teach .claude/skills/teach
$ npx degit toverux/cantrips/skills/tickets .claude/skills/tickets
$ npx degit toverux/cantrips/skills/wait-what .claude/skills/wait-what
$ npx degit toverux/cantrips/skills/writing-for-agents .claude/skills/writing-for-agents
$ curl -fsSL --create-dirs -o .claude/agents/arbiter.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/arbiter.md
$ curl -fsSL --create-dirs -o .claude/agents/deputy.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/deputy.md
$ curl -fsSL --create-dirs -o .claude/agents/conventions.md https://raw.githubusercontent.<redacted>.md
$ curl -fsSL --create-dirs -o .claude/agents/cross-file.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/cross-file.md
$ curl -fsSL --create-dirs -o .claude/agents/design.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/design.md
$ curl -fsSL --create-dirs -o .claude/agents/diff-scan.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/diff-scan.md
$ curl -fsSL --create-dirs -o .claude/agents/efficiency.md https://raw.githubusercontent.<redacted>.md
$ curl -fsSL --create-dirs -o .claude/agents/language-pitfalls.md https://raw.githubusercontent.<redacted>-pitfalls.md
$ curl -fsSL --create-dirs -o .claude/agents/low.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/low.md
$ curl -fsSL --create-dirs -o .claude/agents/removed-behavior.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/removed-behavior.md
$ curl -fsSL --create-dirs -o .claude/agents/reuse.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/reuse.md
$ curl -fsSL --create-dirs -o .claude/agents/simplification.md https://raw.githubusercontent.<redacted>.md
$ curl -fsSL --create-dirs -o .claude/agents/spec-conformance.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/spec-conformance.md
$ curl -fsSL --create-dirs -o .claude/agents/sweep.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/sweep.md
$ curl -fsSL --create-dirs -o .claude/agents/test-value.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/finder/test-value.md
$ curl -fsSL --create-dirs -o .claude/agents/wrappers.md https://raw.githubusercontent.<redacted>.md
$ curl -fsSL --create-dirs -o .claude/agents/verifier.md https://raw.githubusercontent.com/toverux/cantrips/main/agents/verifier.md

This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 16.2k rigs:

{
  "permissions": {
    "deny": [
      "Read(./.env)",
      "Read(./.env.*)",
      "Read(**/.env)",
      "Read(~/.ssh/**)",
      "Bash(git push --force:*)",
      "Read(**/*.pem)",
      "Read(.env)",
      "Read(~/.aws/**)",
      "Read(**/*.key)",
      "Read(**/.env.*)",
      "Bash(git push --force*)",
      "Bash(rm -rf:*)",
      "Bash(rm -rf *)",
      "Bash(sudo *)",
      "Bash(sudo:*)",
      "Bash(rm -rf /)",
      "Bash(git push -f:*)",
      "Read(.env.*)",
      "Bash(git reset --hard*)",
      "Bash(rm -rf /*)"
    ],
    "ask": [
      "Bash(git push:*)",
      "Bash(git push *)",
      "Bash(gh pr merge *)",
      "Bash(git commit:*)",
      "Bash(rm *)",
      "Bash(gh pr merge:*)",
      "Bash(gh pr create *)",
      "Bash(git reset --hard:*)",
      "Bash(npm publish:*)",
      "Bash(git rebase *)"
    ]
  }
}

Skills (26)

Subagents (17)

arbiter
model: fable
/review-gate's arbiter in a run that applies fixes, ruling whether each finding is worth fixing here, and its ruling binds. Not for direct use; it expects the round's message the gate composes.
deputy
model: fable
The session's stand-in, engaged through /deputy, that answers in the user's place every round, gate and closing a skill would put to them. Only the user summons it, through /deputy or by naming it for
conventions
model: opus
A quality finder of /review-gate, hunting through the Conventions lens. Only the user summons it, through /review-gate or by naming it on a diff or a file, and it returns its candidates and their fixe
cross-file
model: opus
A correctness finder of /review-gate, tracing each changed function to its callers and callees. Only the user summons it, through /review-gate or by naming it on a diff or a file, and it returns its c
design
model: opus
A quality finder of /review-gate, hunting through the Design lens. Only the user summons it, through /review-gate or by naming it on a diff or a file, and it returns its candidates and their fixes as
diff-scan
model: opus
A correctness finder of /review-gate, scanning every hunk line by line and the function around it. Only the user summons it, through /review-gate or by naming it on a diff or a file, and it returns it
efficiency
model: sonnet
A quality finder of /review-gate and /simplify, hunting through the Efficiency lens. Only the user summons it, through either skill or by naming it on a diff or a file, and it returns its candidates a
language-pitfalls
model: opus
A correctness finder of /review-gate, hunting the classic pitfalls of the diff's language and framework. Only the user summons it, through /review-gate or by naming it on a diff or a file, and it retu
low
model: opus
/review-gate's one finder at low, hunting the angles and lenses the gate names one at a time. Not for direct use; it expects the message the gate composes.
removed-behavior
model: opus
A correctness finder of /review-gate, auditing what every deleted or replaced line enforced. Only the user summons it, through /review-gate or by naming it on a diff or a file, and it returns its cand
reuse
model: sonnet
A quality finder of /review-gate and /simplify, hunting through the Reuse lens. Only the user summons it, through either skill or by naming it on a diff or a file, and it returns its candidates and th
simplification
model: sonnet
A quality finder of /review-gate and /simplify, hunting through the Simplification lens. Only the user summons it, through either skill or by naming it on a diff or a file, and it returns its candidat
spec-conformance
model: opus
A correctness finder of /review-gate, checking the diff against the spec. Only the user summons it, through /review-gate or by naming it on a diff and a spec, and it returns its candidates as JSON.
sweep
model: opus
/review-gate's sweep at high, a fresh finder over the verified list hunting only defects not on it. Not for direct use; it expects the message the gate composes.
test-value
model: sonnet
A quality finder of /simplify, hunting through the Test-value lens. Only the user summons it, through /simplify or by naming it on a diff or a file, and it returns its candidates and their fixes as JS
wrappers
model: opus
A correctness finder of /review-gate, checking every wrapper or proxy the diff adds or modifies. Only the user summons it, through /review-gate or by naming it on a diff or a file, and it returns its
verifier
model: opus
/review-gate's verifier, judging each candidate on its own evidence as CONFIRMED, PLAUSIBLE or REFUTED with the proving line. Only the user summons it, through /review-gate or by naming it on a claime

Similar rigs

copied ✓