The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data which creates a surface for indirect prompt injection attacks.
Ingestion points: Phase 2 research agents retrieve content from external websites via the WebFetch tool.
Boundary markers: The prompt instructions use standard Markdown headers for structure but do not implement explicit delimiters or 'ignore embedded instructions' warnings for the data retrieved from the web.
Capability inventory: The skill utilizes the Agent tool for parallel execution, Bash for shell access, and Write for file system persistence in Phase 5.
Sanitization: No sanitization, validation, or escaping of fetched web content is performed before it is synthesized into the final report.

deep-dive