Explores intent, requirements, trade-offs, and design via dialogue before planning. Use for new features, components, behaviour changes, ideation, or spec design.
Three-gate Plan -> Execute -> Review workflow for coordinating a coordinator, codex for backend work, and agy for frontend work. Load first for any planning, routing, execution, or review action.
Executes a written plan one phase at a time with Plan → Execute → Review gates. Use when running a plan, advancing the current phase, or continuing implementation in this session.
Use when encountering any bug, test failure, or unexpected behaviour, before proposing fixes. Covers root-cause investigation, hypothesis testing, and fixing at the source.
Use when implementing any feature or bugfix, before writing implementation code. Covers writing the failing test first, watching it fail, then minimal code to pass.
External network access
Connects to servers outside your machine
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.
Multi-model orchestration plugin for Claude Code. The host Coordinator (Claude) plans, routes, reviews, and handles simple edits. Heavier work is dispatched to codex for backend work or Antigravity / agy for frontend work through one MCP tool.
CCG = Claude + Codex + Gemini
Three gates: Plan → Execute → Review. Route each phase by side. Cross-Validation advises unclear or high-impact routing, but never owns implementation.
flowchart TD
A([Request]) --> B{Fuzzy idea or<br/>clear task?}
B -->|"fuzzy idea"| BR["/brainstorm<br/>design doc"]
BR --> WP["/write-plan<br/>PLAN.md + phases"]
B -->|"clear task"| WP
WP --> P["Gate 1 - Plan<br/>frame phase, emit # ROUTE"]
P --> R{Route by side}
R -->|simple| CO["Coordinator edits directly"]
R -->|backend| CX["codex via MCP"]
R -->|frontend| GM["agy via MCP"]
R -->|"full-stack / unclear / high-impact"| CV["Cross-Validation<br/>ask both, reconcile"]
CV --> R
CO --> EX["Gate 2 - Execute<br/>uncommitted work + notes + journal"]
CX --> EX
GM --> EX
EX --> RV{"Gate 3 - Review"}
RV -->|FAIL| P
RV -->|"PASS"| SQ["Phase commit<br/>update .handover.md"]
SQ --> NX{More phases?}
NX -->|yes| P
NX -->|no| V(["verifying-before-completion"])
The canonical spec is skills/coordinating-multi-model-work/SKILL.md — gates, routing table, review semantics, worker contract, and resume artifacts all live there. Everything else in this repo points to it.
User overrides ("use codex", "skip cross-validation", "no external models") always win.
claude plugin marketplace add https://github.com/sitien173/superpowers-ccg
claude plugin install superpowers-ccg
claude --versioncodex --versionagy --versionuv / uvxA single unified server — openmcp — exposes one tool, mcp__plugin_superpowers-ccg_openmcp__run, with a backend field ("codex" or "agy"). It is launched from .mcp.json through uvx at a pinned revision.
Optional defaults pass through from user environment variables. Unset values defer to OpenMCP or backend defaults. Named codex profiles are optional.
| Variable | Purpose |
|---|---|
OPENMCP_AGY_MODEL_DEFAULT | Default model for backend="agy" |
OPENMCP_CODEX_MODEL_DEFAULT | Default model for backend="codex" |
OPENMCP_CODEX_PROFILE_DEFAULT | Default codex profile |
OPENMCP_AGY_REASONING_MODEL | Model used for agy reasoning-mode calls |
OPENMCP_CODEX_REASONING_MODEL | Model used for codex reasoning-mode calls |
OPENMCP_LOG_FILE | OpenMCP log file path (default ~/.openmcp/openmcp.log) |
OPENMCP_LOG_LEVEL | OpenMCP log level (default INFO) |
Slash commands (each loads its shared skill before acting):
/brainstorm — explore intent, requirements, and design via dialogue./write-plan — turn a confirmed design into a phase-based plan./execute-plan — run the active phase under the three gates.Shared skills (namespace superpowers-ccg:):
coordinating-multi-model-work — canonical 3-gate workflow, routing, review, resume artifacts.brainstorming, writing-plans, executing-plans — phase-stage skills loaded by the slash commands.test-driven-development — failing test first, then minimal code (feature/bugfix phases).systematic-debugging — root-cause investigation before any fix.verifying-before-completion — fresh verification evidence before reporting done.Every executable plan lives in docs/plans/<slug>/ (PLAN.md + .handover.md; phase-NN/ folders created lazily). .handover.md is the resume pointer — a new session reads it first, then only the files it lists. Flat plans are documentation only. Bundled worker contracts are read directly from the installed plugin and never copied into consuming repositories.
claude plugin update superpowers-ccg
tests/run.sh
The suite checks hook safety, contract invariants, JSON configuration, version
consistency, and plugin validation when Claude Code is installed. Update the
OpenMCP revision in .mcp.json deliberately, then rerun this suite.
Issues: https://github.com/sitien173/superpowers-ccg/issues
npx claudepluginhub sitien173/superpowers-ccg --plugin superpowers-ccgSuperpowers-CCCG: Skills-driven development workflow with CCCG multi-model orchestration (Claude orchestrator + Codex backend + Cursor DevOps + Gemini frontend)
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows
Use when starting any conversation - establishes how to find and use skills, requiring Skill tool invocation before ANY response including clarifying questions
Agentic development framework for Claude Code — disciplined workflow routing, TDD enforcement, safety hooks, systematic debugging, and code review
24 elite skills across 5 layers: discipline, domain expertise, intelligence, coordination, and execution. Turn Claude Code into a proactive engineering partner.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Evidence-gated AI coding workflow: scan → analyze → plan → TDD → execute → fix → verify → review, powered by Codebase Memory MCP >= 0.9.0 with optional Serena LSP intelligence. Includes blast-radius planning, test/cycle gates, independent review, and Windows Git Bash hook auto-resolution.