claude-tools-runner

A smart tool runner: runs your tests only when files have changed since the last time they were run.

Installation

Stable release

/plugin marketplace add ashleydavis/claude-tools-runner
/plugin install claude-tools-runner@claude-tools-runner

Pre-release / testing (pin to dev branch)

/plugin marketplace add https://github.com/ashleydavis/claude-tools-runner.git#dev
/plugin install claude-tools-runner@claude-tools-runner
/plugin marketplace update claude-tools-runner   # pull latest dev commits

To build and run from a local checkout instead of the marketplace, see DEVELOPMENT.md.

Minimal example

Drop this file at <your-project>/.claude/claude-tools-runner.yaml:

triggers:
  - paths:
      - src/**/*.ts
    commands:
      - run: bun run test
        cooldown: 30s
        timeout: 5m

Now whenever Claude finishes a turn that left changed .ts files under src/ in your working tree, bun run test runs once. The hook records a hash of the matched files; on the next turn, if the same files have not changed and the cooldown (30s) has not elapsed, the command is skipped. If the command runs longer than timeout (5m), it is killed and recorded as a failure.

Let your AI configure this for you

Point your AI assistant at docs/CONFIG-QUICKREF.md and ask it to create or update your claude-tools-runner.yaml. That document is written specifically for AI: concise schema reference, variable table, invocation rules, and ready-to-copy examples with no prose to wade through.

Example prompt to give your AI:

Read docs/CONFIG-QUICKREF.md, then add a trigger to .claude/claude-tools-runner.yaml that runs bun run test whenever a .ts file under src/ changes.

Learn more

• Config guide for AI: Concise reference for AI assistants creating or editing configuration files.
• Configuration: Full configuration reference with detailed explanations.
• How it works: How the plugin works under the hood.
• Development: Instructions for cloning, building, and contributing to the plugin.