The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to install development dependencies using the project's detected package manager (e.g., npm install -D, pnpm add -D, yarn add -D, bun add -d).
[COMMAND_EXECUTION]: A post-installation smoke check is performed using npx eslint --version or npx prettier --version to verify successful setup.
[DYNAMIC_EXECUTION]: The skill writes JavaScript module files (eslint.config.mjs and eslint-plugin-no-emdash.mjs) to the local file system. These files are subsequently executed by the agent platform when running the eslint smoke check. The source code for these files is provided via static templates within the skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository files, including package.json, lockfiles, and existing linting configurations, to determine the appropriate setup path.
Ingestion points: package.json, pnpm-lock.yaml, yarn.lock, bun.lock, package-lock.json, tsconfig.json, and various .eslintrc or prettier.config files.
Boundary markers: None identified.
Capability inventory: Local file writes, package installation, and shell command execution via npx.
Sanitization: The skill uses basic file-existence checks and package name matching to select templates.

setup-toolkit