The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface.
Ingestion points: The skill retrieves data from the web using WebFetch and WebSearch, and from the local filesystem using Read, Glob, and Grep (SKILL.md, Step 1).
Boundary markers: The instructions do not define delimiters or protective prompts to prevent the agent from obeying malicious instructions potentially hidden in analyzed legal documents or web pages.
Capability inventory: The agent has access to Bash and Write tools (SKILL.md frontmatter), which could be misused if the agent is compromised by an indirect injection.
Sanitization: No sanitization or validation of external input is described in the skill's logic.

brief-recon