The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes content from a user-provided directory path, which is a classic surface for indirect prompt injection. Maliciously crafted instructions within the skill being audited could attempt to manipulate the auditing process or the agent's subsequent actions.\n- Ingestion points: Step 1 (Discovery) involves reading the SKILL.md file and inspecting scripts/, references/, and assets/ directories.\n- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to isolate the untrusted data from the agent's reasoning.\n- Capability inventory: The skill possesses the ability to write report files to the local file system (.tmp/scratch/reports/).\n- Sanitization: The workflow does not describe any sanitization, validation, or escaping of the external content before it is processed by the model.

audit-skill-completeness