Ethanon/developer.ai
Lifecycle-shaped Claude Code agent kit: scheduled audits → ready-state issues → developer-agent PRs → multi-persona advisory review. Six named PR reviewers, five audit scanners, one installer. Claude-only by design.
ARCHETYPE
YOLO Cowboy
Permissions? Never heard of 'em. Ships at the speed of `--dangerously-skip-permissions`.
Copy this rig
# review before running: this installs third-party code
$ npx degit Ethanon/developer.ai/.claude ./rig-developer.ai # 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.
$ curl -fsSL --create-dirs -o .claude/agents/class_size_audit.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/class_size_audit.md $ curl -fsSL --create-dirs -o .claude/agents/flaky_test_finder.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/flaky_test_finder.md $ curl -fsSL --create-dirs -o .claude/agents/hanging_refs.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/hanging_refs.md $ curl -fsSL --create-dirs -o .claude/agents/market_watch.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/market_watch.md $ curl -fsSL --create-dirs -o .claude/agents/naming_audit.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/naming_audit.md $ curl -fsSL --create-dirs -o .claude/agents/prompt_audit.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/prompt_audit.md $ curl -fsSL --create-dirs -o .claude/agents/release_audit.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/release_audit.md $ curl -fsSL --create-dirs -o .claude/agents/security_audit.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/audits/security_audit.md $ curl -fsSL --create-dirs -o .claude/agents/audit_groomer.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/backlog/audit_groomer.md $ curl -fsSL --create-dirs -o .claude/agents/feature_agent.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/backlog/feature_agent.md $ curl -fsSL --create-dirs -o .claude/agents/scrum_master.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/backlog/scrum_master.md $ curl -fsSL --create-dirs -o .claude/agents/story_groomer.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/backlog/story_groomer.md $ curl -fsSL --create-dirs -o .claude/agents/installer.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/installer.md $ curl -fsSL --create-dirs -o .claude/agents/alice_security.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/pr-review/alice_security.md $ curl -fsSL --create-dirs -o .claude/agents/bob_engineering.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/pr-review/bob_engineering.md $ curl -fsSL --create-dirs -o .claude/agents/carl_ux.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/pr-review/carl_ux.md $ curl -fsSL --create-dirs -o .claude/agents/gomez_cleancode.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/pr-review/gomez_cleancode.md $ curl -fsSL --create-dirs -o .claude/agents/hyde_blackhat.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/pr-review/hyde_blackhat.md $ curl -fsSL --create-dirs -o .claude/agents/jekyll_whitehat.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/pr-review/jekyll_whitehat.md $ curl -fsSL --create-dirs -o .claude/agents/phil_testing.md https://raw.githubusercontent.com/Ethanon/developer.ai/master/agents/pr-review/phil_testing.md
Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"permissions": {
"deny": [
"Bash(git push *--force*main*)",
"Bash(git push *--force*master*)",
"Bash(git push *-f*main*)",
"Bash(git push *-f*master*)"
]
}
} Subagents (20)
| class_size_audit model: sonnet | Scans the codebase for classes over 1000 lines of executable code (comments, imports, types, markup and stylesheets subtracted) and self-classifies each into auto-accepted, flagged, or investigate. Au |
| flaky_test_finder model: sonnet | Weekly agent that evaluates CI test-run history to distinguish flaky tests (non-deterministic, intermittently failing) from real failures (consistently broken), and scans test source for structural fl |
| hanging_refs model: sonnet | Scans the codebase for dead imports, unused exports, orphan routes, stale env vars, unreferenced Docker services, CSS dead classes, and stale documentation references. Read-only against source; writes |
| market_watch model: sonnet | Weekly read-only agent that surfaces engineering-practice and tech ecosystem shifts at four severity bands (Critical/High/Medium/Low) with an opinionated Recommended action per item. Seeds source list |
| naming_audit model: sonnet | Audits class names in the codebase for suffix/contract mismatches. Reads the Naming Conventions section of docs/ENGINEERING_PRINCIPLES.md as the source of truth, self-classifies findings into `auto-al |
| prompt_audit model: sonnet | (Optional, for projects that ship LLM prompts.) Audits prompt templates for consistency with the project's prompt-rules doc and for drift between code and the prompt-catalog documents. Self-classifies |
| release_audit model: sonnet | Weekly scanner that answers "is this safe to ship?" Combines a deploy-readiness checklist (DB migrations, feature flags, rollback paths, env-var drift) with an Impact × Risk × Effort tech-debt scoring |
| security_audit model: sonnet | Scans the codebase for violations of the rules in engineering/SECURITY_PRINCIPLES.md and the project's own docs/SECURITY.md. Catches routes with no schema validation, identity taken from a request pat |
| audit_groomer model: sonnet | Weekly agent that converts audit-bot findings (security_audit, hanging_refs, naming_audit, release_audit) into pickup-ready GitHub issues. Reads the latest report from each source bot, files one issue |
| feature_agent model: opus | The daily development agent. Owns one unit of work at a time (WIP=1) through a label-driven lifecycle. It drafts a design document first and waits for the owner's `design-approved` label before writin |
| scrum_master model: sonnet | Weekly maintenance agent that fully owns the issue tracker. Auto-closes any open issue whose work has shipped in a merged PR (STRONG or LIKELY match, no human review), auto-creates closed [shipped] tr |
| story_groomer model: sonnet | Daily story-lifecycle agent. Mode A reads approved decision docs (`**Status:** Approved` / `Implemented` / `Landed`), identifies story-shaped H3 sections, suffixes the heading with `[story]`, and file |
| installer model: sonnet | The installer agent walks an adopter through a short Q&A wizard and deploys the developer.ai kit (agents, skills, workflows, templates) into their target repository. Reads the inline tags in each kit |
| alice_security model: sonnet | Alice reviews an open pull request for security concerns introduced by its changes, posting findings as a GitHub PR review. Scoped to the diff plus one-hop neighbors; enforces SECURITY.md and the arch |
| bob_engineering model: sonnet | Bob reviews an open pull request for code quality, posting findings as GitHub PR review comments. Reads the full changed files plus one-hop neighbors for context before commenting, enforces the engine |
| carl_ux model: sonnet | Carl reviews an open pull request for UX and product-feel concerns on the project's frontend. 20 years of phone-game UX experience; scoped to the frontend source folders, user-facing copy, navigation |
| gomez_cleancode model: sonnet | Gomez reviews an open pull request for clean, concise, functional code and for names (methods, variables, classes, files) that communicate intent to a human reading them for the first time. Scoped to |
| hyde_blackhat model: sonnet | Hyde is the blackhat critic. He reads the PR diff plus the Layer-1 reviewers' posted comments (Alice, Bob, Phil primarily; Gomez and Carl when present), then attacks their suggestions: concrete bypass |
| jekyll_whitehat model: sonnet | Jekyll is the whitehat critic. He reads the PR diff plus the Layer-1 reviewers' posted comments (Alice, Bob, Phil primarily; Gomez and Carl when present), then challenges their suggestions from a best |
| phil_testing model: sonnet | Phil reviews an open pull request for unit-test quality, test-first thinking, mocking discipline, failure-mode coverage, and whether method/parameter names reflect how they're meant to be used. ~20 ye |
Permissions
deny (4)
Bash(git push *--force*main*)
Bash(git push *--force*master*)
Bash(git push *-f*main*)
Bash(git push *-f*master*)
ask (0)
—
allow (33)
Bash(*)
Bash(git *)
Bash(gh *)
Bash(npm *)
Bash(npx *)
Bash(node *)
Bash(bun *)
Bash(bunx *)
Bash(pnpm *)
Bash(yarn *)
Bash(tsc *)
Bash(eslint *)
Bash(prettier *)
Bash(docker *)
Bash(docker-compose *)
Bash(jq *)
Bash(curl *)
Bash(curl.exe *)
Bash(ls *)
Bash(ls)
Bash(dir *)
Bash(cat *)
Bash(head *)
Bash(tail *)
Bash(pwd)
Bash(echo *)
Bash(true)
Bash(mkdir *)
Bash(cp *)
Bash(mv *)
Read
Edit(C:\Repos\developer.ai\*)
Write(C:\Repos\developer.ai\*)
Similar rigs
roadhero/claude-code-setup
My actual Claude Code setup: a stack-agnostic engineering spine, auto-activating rule packs, 50 scoped subagents across web/Android/iOS/compute, safety hooks, and a repo scaffolder. From the book "42: The AI Builder's Stack."
Orchestrator 7.3k tok ·
LeeYudok/agents-scaffold
Fork-and-fill bootstrap for AI coding agents (Claude Code first, Codex/Gemini via AGENTS.md) — agents, skills, hooks, tiered rules (P0/P1/P2) and stack presets composed into one enforced pre-commit gate. 한국어/English
Fort Knox 2.3k tok ·
navox-labs/agents
15-agent AI engineering team for Claude Code. Full sprint cycle: Think, Plan, Build, Review, Test, Ship, Reflect. Zero dependencies.
Orchestrator 5.1k tok ·
SeongwoongCho/adaptive-harness
A self-improving harness router for Claude Code.
Orchestrator 2.3k tok ·