Plugin for Claude Code and Codex · v8.1.5

CC Commander picks the right Claude Code tool and shows you the next step.

Type /ccc and say what you need. It reads your project, runs the matching skill, agent or workflow, and ends with a short menu of what to do next.

/plugin install commander --marketplace KevinZai/commander

Older Claude Code, Codex, then first run
# Claude Code before 2.1.292
/plugin marketplace add KevinZai/commander
/plugin install commander

# Codex CLI
codex plugin marketplace add KevinZai/commander
codex plugin add commander@commander-hub

# Then restart and type
/ccc

22,643 down to 7,266

characters of skill listing Claude carries at session start, down 68% with load-on-demand.

Measured on v8.1.5 against v8.1.3. Skill and agent listings together: 7.7k → 3.4k tokens (chars / 3.6 estimate). Method

Numbers reflect v8.1.5, the current marketplace build.

93 skills · 22 agents · 30 hook events · 6 workflows · MIT

What it does

Three jobs, done quietly.

  • 01

    Routes your request

    Describe the job in plain words. Commander matches it to one of 93 skills or 22 specialist agents and runs it. When the match is unclear, it shows the top picks instead of guessing.

  • 02

    Ends every step with a next step

    After a task you get a short picker: the recommended move first, a couple of alternatives, and “just answer”. You click instead of remembering command names.

  • 03

    Loads only what it needs

    32 core skills are visible to Claude. The other 61 open when you type them, so the model keeps its context for your code.

Catalog

Everything it ships, searchable.

Generated from the plugin’s own files at v8.1.5. “Visible to Claude” skills are picked automatically; “type to open” skills run when you type them.

Showing a preview. The full catalog loads with JavaScript.

  • cccvisible to Claude

    Commander home: reads project context, then routes to build, review, ship or the More menu in one or two clicks.

    Try: /ccc · start here

  • ccc-buildvisible to Claude

    Scaffold a new web app, API or CLI from a three-question spec, built by a background agent and verified.

    Try: /ccc-build · build

  • ccc-reviewvisible to Claude

    Code review of the current branch or PR, or a security, performance or whole-project audit, with a severity-rated report.

    Try: /ccc-review · quality

  • ccc-planvisible to Claude

    Turn an idea, bug, refactor or Linear issue into a reviewed plan file with evals before coding.

    Try: /ccc-plan · build

  • ccc-designvisible to Claude

    UI design: landing pages, component systems, polish passes, Figma to code, screenshot to UI.

    Try: /ccc-design · build

  • ccc-systematic-debuggingvisible to Claude

    Root-cause debugging: reproduce, hypothesize, then fix once with a failing test first.

    Try: /ccc-systematic-debugging · quality

  • ccc-fleetvisible to Claude

    Run parallel agents on one task: worktree fan-out, pipeline, review panel, background run or team, then verify.

    Try: /ccc-fleet · orchestration

  • ccc-save-sessionvisible to Claude

    Save the session to a reloadable file: decisions, files, failures and why, exact next step.

    Try: /ccc-save-session · memory

  • ccc-securityvisible to Claude

    Security audit: OWASP Top 10, secrets, dependency CVEs, prompt-injection defense, CodeQL, incident response.

    Try: /ccc-security · security

  • ccc-seovisible to Claude

    SEO router: technical audit and fixes, AI search optimization, content clusters, analytics, programmatic pages, backlinks.

    Try: /ccc-seo · growth

  • revieweragent

    Read-only review of a diff or PR for correctness, security, performance and maintainability. Returns path:line findings and a verdict.

    Try: /ccc-review · quality

  • SessionStarthook

    Replaces 3 separate Node.js spawns with one process that imports each handler's run() function and aggregates their output.

    runs on SessionStart, 5 handlers · lifecycle

How it compares

Same Claude Code, with the workflow layer built.

CapabilityStock Claude CodeWith CC Commander
Picking a toolYou choose the command/ccc reads the project and routes
After each taskA blank promptA next-step picker, recommendation first
Specialist agentsWrite your own22 ready: architect, reviewer, debugger, designer, language reviewers
Lifecycle hooksConfigure yourself30 events, 51 handlers, on by default
Multi-agent workYou write the plan6 workflows with independent verifiers

Comparing skill collections and other plugins? See the full comparison.

Guide

Learn Claude Code properly.

The Claude Code Bible covers setup, prompting, agents, hooks and the Commander Method in one long, searchable read.