The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute the local script scripts/generate-openapi.ts using npx tsx. This executes code within the local project environment to regenerate API documentation.\n- [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface defined by the following characteristics:\n
Ingestion points: The agent is instructed to read and process existing supplement files (e.g., in src/lib/openapi/endpoints/) which contain metadata fields such as description that support markdown and could host hidden malicious instructions.\n
Boundary markers: The agent prompt templates do not specify any delimiters or warnings to ignore instructions embedded within the data it reads.\n
Capability inventory: The agent has the ability to write to the local filesystem and execute commands, providing a path for an injection to cause unauthorized changes or actions.\n
Sanitization: The instructions do not include any steps for sanitizing or validating the input from these files before processing them.

openapi-endpoints