The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Documentation content is retrieved from the colinhacks/zod repository on GitHub using curl and tar commands in the sync-docs.sh script. This is standard behavior for maintaining a local documentation mirror and targets a well-known project source.
[REMOTE_CODE_EXECUTION]: The script scripts/sync-docs.sh contains a command that pipes data from a GitHub API call into the Python interpreter using the -c flag. While this is used for data extraction from a JSON response rather than executing remote script content, it is noted as a network-dependent execution pattern.
[PROMPT_INJECTION]: The skill ingests external documentation from a remote repository, which theoretically introduces a surface for indirect prompt injection if the source content were ever compromised. -- Ingestion points: Markdown and MDX files in the docs/ directory. -- Boundary markers: Absent; the agent is directed to treat the vendored content as an authoritative reference. -- Capability inventory: The skill does not possess high-risk runtime capabilities like eval() or arbitrary subprocess calls that act upon the ingested data. -- Sanitization: Content is mirrored directly from the source repository without automated filtering.

zod