Claude Code plugin

Waggle for Claude Code

Run Waggle's evidence-backed code audits from inside Claude Code. Every finding is checked against your code before you see it, and fixes happen on a branch, with your approval.

Needs Claude Code, Node.js 22.7 or newer, and git. Runs on your machine with your own Claude account.

  1. Step 1

    Add the plugin

    In a Claude Code session, run these two commands. The first adds AI Theresa's plugin marketplace; the second opens the plugin's details, where you choose Install for you.

    /plugin marketplace add ai-theresa-lab/ai-theresa-waggle-skill
/plugin install theresa-waggle@ai-theresa

  2. Step 2

    Or install from a terminal

    The same two steps from your shell. The plugin loads the next time you start Claude Code, or right away if you run /reload-plugins in a session that is already open.

    claude plugin marketplace add ai-theresa-lab/ai-theresa-waggle-skill
claude plugin install theresa-waggle@ai-theresa

  3. Step 3

    Check that it works

    Type /theresa-waggle and its four commands appear. Then run the free report command below: the first command installs the Waggle engine on your machine, which takes about a minute, once.

    /theresa-waggle:report

What you can run

  • /theresa-waggle:ask

    Answers one question about your code, with file and line evidence. A few minutes.

    Typical cost: $0.30–1

  • /theresa-waggle:scan

    A full audit: security, architecture, release engineering, business logic and more. Usually 20–60 minutes, in the background.

    Typical cost: $5–30, capped

  • /theresa-waggle:report

    Shows a running scan's progress, or a finished one's findings and reports.

    Typical cost: Free

  • /theresa-waggle:fix

    Re-checks each finding against your current code and fixes the confirmed ones on a new branch, one commit each.

    Typical cost: Your usual Claude Code use

Scans and fixes start only when you type the command, and every scan shows its cost and cap first. Re-scanning the same code is incremental and reports what is fixed, new or still open.

Good to know

  • Everything runs on your machine. Your code, questions and reports stay there and go only to the model providers you use.
  • Model calls use your Claude subscription login or your ANTHROPIC_API_KEY. If OPENAI_API_KEY is set, report writing uses it first.
  • Scans are read-only. Only the fix command changes code, on a separate branch, and it never pushes.
  • Anonymous usage telemetry is on by default and never includes code, paths or findings. Ask Claude to turn off Waggle telemetry, or set THERESA_TELEMETRY=0.
  • Updating: run /plugin, open Marketplaces, select ai-theresa and choose Update marketplace (or, from your shell, claude plugin marketplace update ai-theresa and then claude plugin update theresa-waggle@ai-theresa). Updates are not automatic for marketplaces outside Anthropic's unless you enable auto-update there.
Plugin on GitHubThe open-source engineAbout Waggle