The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to fetch metadata and validate packages.
Evidence includes shell command patterns like curl https://registry.npmjs.org/{package}/latest | jq and pnpm validate:packages.
The package name {package} is interpolated directly into shell strings. If this variable is sourced from untrusted file content without proper escaping, it could lead to command injection (e.g., a package name like pkg; rm -rf /).
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves reading and acting upon content from rule files located in the repository.
Ingestion points: The agent reads content from packages/content/rules/en/{category}/{slug}.mdx to identify packages and keywords.
Boundary markers: No specific delimiters or safety instructions are provided to the agent to ignore potentially malicious instructions embedded within these files.
Capability inventory: The agent has the ability to execute network requests via curl, run project scripts via pnpm, and write modifications back to the rule files.
Sanitization: While the instructions require verifying that packages exist on the npm registry, there are no explicit instructions for sanitizing or escaping the strings extracted from rule files before they are used in shell commands or interpolated into prompts.

find-npm-packages