~ / rigs / kchojn / claude-code-tradecraft

kchojn/claude-code-tradecraft

Claude Code plugin: a delegation-first lead, a two-stage PR review, a review watcher, and the sourced rules behind them.

↗ GitHub ★ 1 MIT updated today project Claude Code Claude plugin
share on X
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
CONTEXT TAX · EVERY TURN
~2.1k tokens
Moderate · median rig: 2.3k · breakdown
GUARDRAILS
0/5
No committed guardrails · details
Quality 0/100 Guardrails v2

Copy this rig

# review before running: this installs third-party code
$ npx degit kchojn/claude-code-tradecraft/agents ./rig-claude-code-tradecraft/agents
$ npx degit kchojn/claude-code-tradecraft/skills ./rig-claude-code-tradecraft/skills

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 kchojn/claude-code-tradecraft/skills/lead .claude/skills/lead
$ npx degit kchojn/claude-code-tradecraft/skills/pr-review-watch .claude/skills/pr-review-watch
$ npx degit kchojn/claude-code-tradecraft/skills/pr-review .claude/skills/pr-review
$ curl -fsSL --create-dirs -o .claude/agents/deep-reasoner.md https://raw.githubusercontent.com/kchojn/claude-code-tradecraft/main/agents/deep-reasoner.md
$ curl -fsSL --create-dirs -o .claude/agents/implementer.md https://raw.githubusercontent.com/kchojn/claude-code-tradecraft/main/agents/implementer.md
$ curl -fsSL --create-dirs -o .claude/agents/oracle.md https://raw.githubusercontent.com/kchojn/claude-code-tradecraft/main/agents/oracle.md
$ curl -fsSL --create-dirs -o .claude/agents/qa-runner.md https://raw.githubusercontent.com/kchojn/claude-code-tradecraft/main/agents/qa-runner.md
$ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.com/kchojn/claude-code-tradecraft/main/agents/reviewer.md
$ curl -fsSL --create-dirs -o .claude/agents/scout.md https://raw.githubusercontent.com/kchojn/claude-code-tradecraft/main/agents/scout.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 (3)

Subagents (6)

deep-reasoner
model: opus
Use for design, hard debugging, algorithm design, and verifying work with many hidden edge cases by reading code. Returns a conclusion, the key risks, and a concrete recommendation. Never for manual t
implementer
model: sonnet
Use for writing phases with a clear spec - feature code, bug fixes whose cause is already established, refactors. Implements the brief, runs the existing checks, and returns what changed and how it wa
oracle
model: fable
Use for one read-only decision call - the plan for a major architecture, migration or performance change, a question deep-reasoner could not settle, a phase that failed twice, or a next phase the lead
qa-runner
model: haiku
Use for manual and end-to-end testing - launch the app, click through flows in a browser or desktop build, take screenshots, and report pass/fail against the brief's steps. Never edits code.
reviewer
model: sonnet
Use for the find and verify phases of a code review. Read-only; reads code, diffs, history and PR metadata, and returns candidates or verdicts with file:line evidence. Never edits or posts.
scout
model: haiku
Use for read-only lookups - finding files and symbols, reading logs, running existing tests or builds and reporting the result. Returns locations and short evidence, not file dumps. Never writes.

Similar rigs

copied ✓