claude-chat-namer

Claude Code plugin that names, searches, and helps you resume your conversations. No more UUID chaos.

Listed in Anthropic's plugin directory.

Claude Code stores conversations as UUID-named .jsonl files. This plugin names them, searches everything you typed across all your projects, gives you a claude --resume command that works from anywhere, and can pull an old conversation's context into the one you are in.

Works wherever Claude Code keeps its transcripts on disk: the terminal, the desktop app and the IDE extensions. In claude.ai chats there are no local transcripts to read.

Install

As Claude Code Plugin (recommended)

The repo is its own plugin marketplace. Add it once, then install the plugin:

claude plugin marketplace add Jamaalx/claude-chat-namer
claude plugin install chat-namer@claude-chat-namer

Or from inside Claude Code:

/plugin marketplace add Jamaalx/claude-chat-namer
/plugin install chat-namer@claude-chat-namer

Restart Claude Code (or run /reload-plugins) and the /chat-namer:* commands and the auto-naming hook are active. Update later with claude plugin update chat-namer@claude-chat-namer.

Try it without installing

git clone https://github.com/Jamaalx/claude-chat-namer.git
claude --plugin-dir ./claude-chat-namer

As standalone CLI

npx claude-chat-namer

Or install globally:

npm install -g claude-chat-namer
claude-chat-namer list

Features

Real names, from wherever they come from

Every conversation gets the best name available, tagged with its source:

A Stop hook also gives each new conversation a first-message name after its first exchange, as a fallback. No configuration needed.

Search everything you typed, in every project

> /chat-namer:search stripe webhook

2 result(s) for "stripe webhook":

  1. [renamed] Stripe billing
     Oct 28, 09:12 PM | 214 msgs | 1890 KB | /home/me/shop | eeee5555-...
  2. [ai] Checkout refactor
     Oct 12, 04:40 PM | 96 msgs | 702 KB | /home/me/api | 1a2b3c4d-...
     "...then verify the stripe webhook signature before we mark the order paid..."

Every word must match, in any order. Title hits come first; content hits show where they matched. Subagent traffic and tool output are left out.

Resume from anywhere

claude --resume <uuid> only finds a conversation from the directory it was started in. chat-namer gives you the full command:

> /chat-namer:resume stripe

  Stripe billing

  Run in a new terminal:
    cd /home/me/shop && claude --resume eeee5555-5555-4555-8555-555555555555

From a terminal, claude-chat-namer resume stripe launches it directly.

Recap: continue an old thread here

> /chat-namer:recap checkout refactor

Claude reads the last exchanges of that conversation (your messages and its answers, no tool output) and tells you where it stopped, so you can carry on in the current session without switching.

Slash Commands (inside Claude Code)

CLI Usage

claude-chat-namer                          # Interactive mode (browse + rename)
claude-chat-namer list [--all]             # List conversations
claude-chat-namer search stripe --all      # Search titles and content
claude-chat-namer resume stripe --all      # Jump back in, in the right directory
claude-chat-namer recap stripe --all       # Print the last messages
claude-chat-namer name c348 "Wolt Automation"  # Name by UUID prefix
claude-chat-namer unname c348              # Remove a name
claude-chat-namer auto                     # Auto-name all unnamed
claude-chat-namer help                     # All commands and options

Options: --all, --project <name>, --limit <n>, --json (machine-readable, for scripts and agents), --print (resume: only print), --turns <n> (recap).

Fast on big histories

Transcripts are streamed, never loaded whole, and the metadata is cached. Transcripts are append-only, so a grown file is read only from where the last run stopped. On 8 GB of transcripts (577 conversations): about 4 s the first time, about 0.3 s after that.

How it Works

1. Scans ~/.claude/projects/ (or $CLAUDE_CONFIG_DIR/projects/) for conversation .jsonl files
2. Reads titles, the directory, and what you typed from each transcript
3. Stores your names in a chat-names.json file (per project directory) and a speed cache in ~/.claude/chat-namer/
4. Never modifies conversation files and makes no network requests. See PRIVACY.md

Plugin Structure

claude-chat-namer/
├── .claude-plugin/
│   ├── plugin.json           # Plugin manifest
│   └── marketplace.json      # Marketplace manifest (lets you `claude plugin install` this repo)
├── assets/icon.svg           # Directory icon
├── skills/
│   ├── list/SKILL.md         # /chat-namer:list
│   ├── search/SKILL.md       # /chat-namer:search
│   ├── rename/SKILL.md       # /chat-namer:rename
│   ├── resume/SKILL.md       # /chat-namer:resume
│   ├── recap/SKILL.md        # /chat-namer:recap
│   └── auto-name/SKILL.md    # /chat-namer:auto-name
├── hooks/
│   └── hooks.json            # Stop hook for auto-naming
├── scripts/
│   ├── index.js              # Core API
│   ├── cli.js                # CLI interface
│   └── hook-auto-name.js     # Hook script
├── test/
│   └── smoke.sh              # End-to-end test against a throwaway HOME
├── package.json
└── README.md

Development

bash test/smoke.sh          # runs the CLI + hook against fake transcripts
claude plugin validate .    # checks plugin.json / marketplace.json

CI runs the smoke test on Node 18, 20 and 22. See CHANGELOG.md for release notes.

Zero Dependencies

Uses only Node.js built-ins (fs, path, os, readline). No npm install needed.

License

MIT