By nuttaruj
Coordinated software-house team for AI coding CLIs — 18 specialist agents, Core 10 phase skills (1 router + 9 phase), 6-phase lifecycle. Pre-commit gates (S1-S5 simplicity / T1-T6 testing / F1-F5 failure-mode) catch bugs before ship. On Claude Code, ships as a pure marketplace plugin (agents / skills / commands / hooks) — the always-on judgment core is delivered by a SessionStart hook, so install and uninstall touch no global config. See https://github.com/nuttaruj/rolepod.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
AI/ML Engineer specializing in LLM integration, RAG systems, prompt engineering, agent design, embeddings, and Anthropic/OpenAI API usage. Distinct from data-scientist (statistics) — focus is applied AI features in production code.
Backend Specialist. Builds APIs, business logic, database models, integrations. Excludes specialist domains (billing/AI/data analytics) which have dedicated agents.
FinTech / Monetization Engineer. Owns billing, payments, credits, subscriptions, financial data integrity. Path-scoped to billing/payments/credits modules.
Business Strategist for pricing models, cost/ROI analysis, financial modeling, competitor research. Commercial layer — distinct from product-manager (feature decisions).
Customer Success — user onboarding, FAQ, support content, technical-to-user translation. Distinct from tech-writer (internal docs) and growth-marketer (acquisition).
Use after a change is made and before claiming the work is done — prove it with evidence (tests, build, typecheck, curl, logs, screenshot, browser). State limitations explicitly when verification is not possible. Phase = Verify.
Use when something is broken — error appears, test fails, build breaks, output is wrong, regression returns. Reproduce, trace upstream to root cause, write a failing test, ship a minimal fix. Phase = Build / Debug.
Use at the end of a development branch — pre-merge gate, CI lane discipline, 4-option finish menu (merge, PR, keep open, discard), release checklist for production launches. Phase = Ship.
Use when executing an approved plan or a clear single-file edit — TDD for risky paths, surgical edits, bounded delegation, worktrees only when real filesystem isolation is needed. Phase = Build.
Use when the session is long, the repo is unfamiliar, the work is multi-file, you are stuck, or you need to escalate to a stronger model. Context budget, session hygiene, zoom-out, deep triage, escalate, onboarding. Phase = Recovery / Re-context / Escalate.
Uses power tools
Uses power tools
Uses Bash, Write, or Edit tools
Uses Bash, Write, or Edit tools
Rolepod turns Claude Code, Codex CLI, Gemini CLI, and Cursor IDE into a disciplined software-house team — a workflow router, 18 specialist agents, and gates that catch bugs before they reach a commit.
It is one source of truth rendered into a native plugin for each CLI. No CLI is the "default" — all four are first-class. Rolepod carries zero project-specific configuration, so it works in any repository from the first session.
Rolepod starts the moment you give your CLI a task — you don't run a command or pick a mode. A router skill reads the request and places it in the workflow: a one-line typo fix goes straight to the edit; a vague "build me X" gets pulled back into a spec conversation first.
Every real change then moves through six phases:
Define → Plan → Build → Verify → Review → Ship
Each phase has one skill that runs it, and each skill pulls in specialist agents when the work needs depth. Before any commit, gates fire automatically — simplicity, tests, failure-mode — as soft reminders on ordinary code and hard blocks on high-risk paths like auth, billing, and migrations.
You invoke nothing for this; it just happens. For a deliberate run through every phase with no skips, invoke /rolepod-full.
write-spec. Turns a fuzzy request into a spec, shown back in chunks short enough to actually read and approve.write-plan. Breaks the spec into tasks, assigns agent ownership, writes a cohesion contract before any parallel work.implement-plan. Executes the plan test-first with bounded delegation. Bug fixes take the debug-issue path: reproduce → failing test → minimal fix.check-work. Proves the change with evidence — tests, build, curl, a screenshot — never just a "done".review-code. Multi-axis review, with adversarial pressure on high-risk diffs.finish-work. Pre-merge gate, CI lanes, and a 4-option finish menu.Two skills run across phases: simplify-code (behavior-preserving cleanup) and manage-context (recovery when a session is long, stuck, or in an unfamiliar repo).
Pick your CLI. Rolepod installs only itself — agents, skills, hooks, manifests. No third-party tools are installed for you.
# Install
claude plugin marketplace add nuttaruj/rolepod
claude plugin install rolepod@rolepod
# Update
claude plugin marketplace update rolepod
claude plugin update rolepod@rolepod
# Uninstall
curl -fsSL https://raw.githubusercontent.com/nuttaruj/rolepod/main/bootstrap.sh | bash -s -- --uninstall --target=claude
# Install — the plugin carries skills + hooks; the 18 agents need the installer
codex plugin marketplace add nuttaruj/rolepod
codex plugin install rolepod@rolepod
curl -fsSL https://raw.githubusercontent.com/nuttaruj/rolepod/main/bootstrap.sh | bash -s -- --target=codex
# Update
codex plugin marketplace upgrade rolepod
codex plugin remove rolepod@rolepod && codex plugin add rolepod@rolepod
# Uninstall
curl -fsSL https://raw.githubusercontent.com/nuttaruj/rolepod/main/bootstrap.sh | bash -s -- --uninstall --target=codex
Codex hooks are registered but inert until you opt in — run codex features enable plugin_hooks. Agents, skills, and the AGENTS.md gate rules work without it.
# Install
curl -fsSL https://raw.githubusercontent.com/nuttaruj/rolepod/main/bootstrap.sh | bash -s -- --target=gemini
# Update — re-run with --force
curl -fsSL https://raw.githubusercontent.com/nuttaruj/rolepod/main/bootstrap.sh | bash -s -- --target=gemini --force
# Uninstall
curl -fsSL https://raw.githubusercontent.com/nuttaruj/rolepod/main/bootstrap.sh | bash -s -- --uninstall --target=gemini
Recommended — native team marketplace (no terminal, no ~/.cursor/plugins/local/ clutter). Cursor reads the committed .cursor-plugin/marketplace.json at the repo root and lists rolepod in the IDE plugin panel for one-click install / update / uninstall:
https://github.com/nuttaruj/rolepodrolepod entrynpx claudepluginhub nuttaruj/rolepod --plugin rolepodWordPress operations toolkit for Claude Code / Cursor / Codex / Gemini — wp-cli + REST + scoped fs + 89+ MCP tools, opt-in the rolepod-wp WordPress plugin unlocks execute-php + runtime introspection + one-click pair setup. Extension Protocol v1 — runs as a child of the rolepod parent (v2.7+) when its `<git-root>/.rolepod/parent-active` marker is present, standalone otherwise.
Multi-platform UI/mobile automation for AI agents — 8 shipped skills (verify-ui, check-errors, audit-a11y, visual-diff, scaffold-e2e, measure-cwv, audit-page-budget, audit-seo) + MCP server with 30 tools. Works standalone OR with the `rolepod` parent plugin: when the marker file `<git-root>/.rolepod/parent-active` is present (written by the parent's SessionStart hook), evidence routes to `<git-root>/.rolepod/evidence/` with a `manifest.json` per Extension Protocol v1, so parent's `check-work` skill can aggregate UI verify results into its phase report. v0.5 completed the UI verification surface (console + network observability, hover/drag/fill_form/upload/dialog, runtime emulation, multi-page, gated JS eval).
Data-layer safety + truth verifier for AI agents — MCP server with 5 rolepod_db_* tools + 5 shipped skills (db-introspect, db-query, db-explain, db-migrate-verify, db-write). Works standalone OR with the `rolepod` parent plugin: when the marker file `<git-root>/.rolepod/parent-active` is present (written by the parent's SessionStart hook), evidence routes to `<git-root>/.rolepod/evidence/` with a `manifest.json` per Extension Protocol v1, so the parent's check-work skill can aggregate DB state as a first-class evidence type. Not a generic SQL client — the value is safe mutation (preview→confirm→commit/rollback), schema-drift detection, and DB state as PASS/FAIL evidence.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
Comprehensive PR review agents specializing in comments, tests, error handling, type design, code quality, and code simplification
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques