71
plugin skills
// oss · claude code plugin · free for now
CC Commander.
The AI project manager that sits above your Claude Code sessions.
Not another prompt pack. CC Commander orchestrates 71 plugin skills across 22 specialist agents and 23 lifecycle hooks, compounds knowledge from every session into the next, and ships real code while you sleep. One install. Zero config. Free for now.
22
specialist agents
23
lifecycle hooks
free
for now
what it does
One command turns Claude Code into a staffed dev team.
Most Claude Code tools live inside your session — they inject prompts, add context, append rules. CC Commander runs a level above all of that: it picks which skill fires, routes complexity to the right specialist, and tracks progress across builds.
[01]
Click-First Workflows
13 /ccc-* specialist workflows — start, build, review, ship, plan, design — surfaced as clickable chips. You click; it dispatches.
[05]
The Fable Method
12 enforceable gates distilled from Claude Fable 5 — never trust a single pass, prove before you alarm, context is disposable — so any model running CCC produces Fable-shaped results. Toggle with
/ccc-fable.[02]
22 Specialist Agents
Sub-agent personas — architect, reviewer, security-auditor, debugger and more — each tuned to a role, with 23 lifecycle hooks wiring the session end to end.
[03]
Vendor Aggregation
Aggregates 18 vendor packages — gstack, Superpowers, Compound Engineering and others — into one interface. 457+ total skills, auto-detected and sequenced.
[04]
Model-Tier Routing
Auto-selects the reasoning tier per task — Haiku 4.5 for bulk ops, Sonnet 5 for shipping, Opus 4.8 for everyday orchestration, Fable 5 for deep reasoning. No manual model switching.
how it works
You type. It plans. Agents build.
01 · one command
Auto-detect
- npx / plugin install
- reads package.json
- picks the profile
→
02 · you ask
It plans
- /ccc menu
- spec interview
- parallel dispatch
→
03 · you ship
It learns
- tests + types + e2e
- lessons captured
- skills compound
▸ why it matters — one command turns Claude Code into a staffed dev team: 71 skills, 22 agents, zero config.
in production
CC Commander grew from a single frustration: the Claude Code ecosystem had fragmented into dozens of separate prompt packs, extensions, and skill bundles — each configured by hand, none aware of the others. Every new tool meant another setup cycle, and none of them carried what you learned in the last session into the next. Nobody had stitched the ecosystem into one interface.
Commander does. It aggregates 18 vendor packages — gstack, Superpowers, Compound Engineering and 15 others — into 457+ total skills surfaced through one click-first menu. Auto-detect what you have installed, sequence the tools in the right order, route each task to the right model tier, and inject past lessons into every new session. Ships under MIT license with a single npm dependency (figlet) and a clean security audit — 0 issues.
v6.4.0’s headline is the Fable Method — the operating doctrine distilled from Claude Fable 5, encoded as 12 enforceable gates (never trust a single pass, loops need verifiers/state/stops, prove before you alarm, context is disposable) so any model running CCC — Sonnet, GPT, anything — produces Fable-shaped results. The method IS the moat. Run /ccc-fable to turn it on, check status, or run audit to self-check which gates the session is violating.
v6.3.0’s Orchestrator/Executor doctrine still ships — “pay for Fable on the thinking, not the typing.” A high-reasoning orchestrator (Claude Fable 5 or Opus 4.8) writes the plan; a fast executor (GPT-5.5 via codex, or a Sonnet subagent) implements it; the orchestrator verifies against acceptance criteria. Roughly 70% token savings on large tasks. Run /ccc-orchestrate to use it, /ccc-handoff to reset context before quality degrades, and /ccc-adopt to bring the doctrine into any existing project.
Install the AI layer that manages your AI.
OSS · free for now · one install.