Command Catalog

Generate a report of the slash commands you can actually use in your current environment (project + plugins + MCPs), so you can escape the “discoverability crisis” caused by too many tools.

Languages

English | 简体中文 | 繁體中文 | 日本語 | Español | Français | Deutsch | Português(BR) | Italiano | Türkçe | Bahasa Indonesia | Tiếng Việt | ไทย | Polski | Nederlands | Svenska | Русский | Українська | 한국어 | العربية | עברית | हिन्दी

When to use it

• You keep forgetting what /commands exist in a project.
• You want a quick “capabilities manifest” an agent can read without guessing.
• You suspect shadowing/duplication (same name defined in multiple places).

Quickstart (Claude Code plugin)

Run these commands in the Claude Code chat input (paste + send).

Add this repository as a plugin marketplace:

/plugin marketplace add PJH720/slash-command-catalog

Install the plugin from that marketplace:

/plugin install command-catalog@pjh720-slash-command-catalog

/reload-plugins

Then run it:

/command-catalog

Outputs

By default, the command writes these files to the current working directory:

• Human-friendly: ./CommandCatalog.html (single searchable page)
• Agent-friendly: ./llms.txt and ./llms-full.txt (no HTML parsing required)

Optional (if enabled by your environment/plugin setup):

• Human-friendly (expanded): ./llm-full.html

What it scans

• .claude/skills/ (SKILL.md)
• .claude/commands/ (markdown command files)
• .claude/plugins/ (plugin configs, if present)
• .mcp.json (MCP servers summary, if present)

Notes

• The runner is zero-install (pure Node.js ESM) and can also be executed directly:
  • node /absolute/path/to/scripts/scan_and_report.js
• If your Claude Code environment blocks command execution, you may need to approve the node .../scan_and_report.js command when prompted.