The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources such as OpenAPI specifications, GraphQL schemas, and source code annotations to generate documentation. While these sources could theoretically contain malicious instructions, the risk is mitigated by the skill's restricted toolset.
Ingestion points: Processes external schema files, tests, and code artifacts as defined in Step 3 of the instructions.
Boundary markers: The skill does not define explicit boundary markers or specific instructions to ignore commands that might be embedded within the source artifacts.
Capability inventory: The skill is limited to Read, Write, Edit, Glob, and Grep tools, which are necessary for documentation tasks but do not grant network access or the ability to execute code.
Sanitization: There are no explicit instructions for sanitizing content extracted from schemas before its inclusion in the generated documentation output.

api-documentation