The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data which could contain malicious instructions. Ingestion points: The skill reads from transcript_path and all files within outputs_dir (referenced in SKILL.md, Step 1). Boundary markers: No explicit delimiters or instructions are provided to the agent to disregard embedded instructions within the processed files. Capability inventory: The skill utilizes the Read, Grep, Glob, and Write tools to perform analysis and save a grading.json report to the filesystem (SKILL.md, Step 5). Sanitization: No sanitization, escaping, or schema validation of the external content is performed prior to the agent evaluating it against expectations.

skill-grader