The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from public package registries (npm and PyPI) to analyze package metadata for suspicious signals.
Ingestion points: Remote metadata is retrieved in scripts/agent.py via query_pypi_package and query_npm_package functions.
Boundary markers: The skill does not implement explicit boundary markers or delimiters for the ingested metadata.
Capability inventory: The skill performs network GET requests to well-known registry domains, reads local dependency files (e.g., package.json, requirements.txt), and writes analysis reports to the local filesystem.
Sanitization: External data is parsed as structured JSON and used strictly for heuristic scoring and comparison logic, minimizing the surface for instruction injection.
[DATA_EXFILTRATION]: The skill communicates with legitimate, well-known services including pypi.org, npmjs.org, and pypistats.org to retrieve package information and download statistics. These network operations are essential for the skill's stated purpose of supply chain security auditing and do not target unauthorized or suspicious domains.

detecting-typosquatting-packages-in-npm-pypi