# ccdesk

> ccdesk is a local web app, started with `npx ccdesk`, that shows and edits everything that shapes Claude Code on one machine: instruction files (CLAUDE.md, AGENTS.md, rules and imports), settings at every scope, permission rules, hooks, MCP servers, subagents, skills, plugins, output styles and auto memory. For the directory it is started in, it lists each piece in the order Claude Code loads it and names the file it comes from. Every edit is shown as a diff first and backed up, and can be undone from Recent changes. It listens on 127.0.0.1 only and sends nothing off the machine.

## Start

- [Getting started](https://ayagmar.github.io/ccdesk/getting-started.md): Run ccdesk, what it reads and writes, and its command-line options.
- [Reading the interface](https://ayagmar.github.io/ccdesk/interface.md): The route strip, the timetables, the inspector, filters, the command palette and keyboard shortcuts.

## What ccdesk shows and changes

- [Instructions](https://ayagmar.github.io/ccdesk/instructions.md): Which CLAUDE.md, AGENTS.md and rule files load in a directory, in what order, and why a file is skipped.
- [What Claude sees](https://ayagmar.github.io/ccdesk/what-claude-sees.md): What your config puts into context when a session starts, with sizes and estimated tokens, and what joins when Claude touches a file.
- [Settings](https://ayagmar.github.io/ccdesk/settings.md): The value Claude Code uses for each setting, the file it comes from, and how to change it safely.
- [Permissions](https://ayagmar.github.io/ccdesk/permissions.md): Every allow, ask and deny rule in force, rules ccdesk suggests from prompts you answered, and what each built-in tool does here.
- [Tool usage counts](https://ayagmar.github.io/ccdesk/tool-usage.md): How ccdesk counts tool calls and your answers to permission prompts from Claude Code's transcripts, and what reaches the page.
- [Hooks](https://ayagmar.github.io/ccdesk/hooks.md): The hooks that run in a directory, which ones are switched off, and how to add, change or remove one.
- [MCP servers](https://ayagmar.github.io/ccdesk/mcp-servers.md): Which MCP servers a session here starts, which wait for approval or are switched off, and how ccdesk changes them.
- [Subagents](https://ayagmar.github.io/ccdesk/subagents.md): The subagents available in a directory, which definition wins, and how to create, change or delete an agent file.
- [Skills, commands, output styles and plugins](https://ayagmar.github.io/ccdesk/skills-and-plugins.md): Where ccdesk finds skills, slash commands, output styles and plugins, and which ones are in force.
- [Memory](https://ayagmar.github.io/ccdesk/memory.md): The auto memory Claude Code saved for a directory, the checks ccdesk runs on it, and how to delete a memory or fix MEMORY.md.
- [Recent changes, backups and undo](https://ayagmar.github.io/ccdesk/recent-changes.md): Where ccdesk keeps backups, how undo and restore work, and when a restore is refused.

## Reference

- [Safety and privacy](https://ayagmar.github.io/ccdesk/safety-and-privacy.md): What ccdesk reads, what it never reads or writes, how secrets are hidden, and how the local server is protected.
- [Platform notes](https://ayagmar.github.io/ccdesk/platforms.md): What differs on Linux, macOS and Windows.
- [FAQ](https://ayagmar.github.io/ccdesk/faq.md): Short answers to common questions about ccdesk.
- [Contributing](https://ayagmar.github.io/ccdesk/contributing.md): The development loop, the generated Claude Code data, the docs site, and how releases are made.

## Optional

- [All docs in one file](https://ayagmar.github.io/ccdesk/llms-full.txt)
- [Source code](https://github.com/ayagmar/ccdesk)