The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes user-supplied specification files or JSON data to determine its research path.
Ingestion points: The skill accepts a <spec file | spec JSON> as its primary input argument.
Boundary markers: There are no instructions provided to the agent to treat the contents of the spec file as data rather than instructions, nor are there any delimiters defined to separate user input from the skill's logic.
Capability inventory: The skill is authorized to use Read, Grep, Find, and Ls tools, which allow for broad inspection of the local file system.
Sanitization: No sanitization, validation, or escaping mechanisms are described for the input data before it influences the agent's behavior and search queries.

clotho-research