The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from external and project-provided sources, such as npm audit logs and registry metadata, which could be exploited if those sources contain malicious strings.
Ingestion points: Processes data from package.json (specifically the overrides section), npm audit --json output, and metadata fetched from the npm registry via npm view.
Boundary markers: The skill does not define explicit delimiters or instructions to the agent to ignore embedded commands within the ingested JSON data.
Capability inventory: The skill has broad capabilities including executing package installations (npm install), running project scripts (npm run build, npm test), and committing changes to the git repository.
Sanitization: There is no explicit requirement in the instructions to sanitize or validate package names, versions, or other metadata before interpolating them into shell commands.
[COMMAND_EXECUTION]: The skill uses various shell commands to manage dependencies and project state.
Evidence: Executes npm install, npm audit fix, npm run build, npm test, and several git commands.
[EXTERNAL_DOWNLOADS]: The skill interacts with the official npm registry to fetch metadata and update packages.
Evidence: Queries registry.npmjs.org using npm view and downloads updated dependencies using npm install.

patch-vulnerabilities