The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes *-audit-results.json files from the current working directory to generate recommendations and interactive walk-through steps. Maliciously crafted data in these files (specifically in the finding or recommendation fields) could potentially influence the agent's behavior.
Ingestion points: SKILL.md Phase 1 (reads *-audit-results.json from the current working directory).
Boundary markers: None present to distinguish untrusted data from instructions.
Capability inventory: The skill uses Bash, Read, Write, and AskUserQuestion tools, allowing for command execution and file system modification.
Sanitization: The skill performs structural JSON validation and schema checks but does not sanitize the natural language content within the JSON fields used for coaching.

ads-next