The Agent Skills Directory

Prompt Injection (LOW): The skill has a significant indirect prompt injection surface due to its core functionality of scraping and searching the web.
Ingestion points: Untrusted data enters the agent context through mcp__perplexity__search, mcp__context7__query-docs, and mcp__firecrawl__firecrawl_scrape (found in SKILL.md).
Boundary markers: The instructions lack explicit delimiters or safety warnings to ignore instructions embedded within the research results.
Capability inventory: The skill is granted Write permissions to the file system and access to several external tool APIs.
Sanitization: There is no mention of sanitizing or validating external content before it is used to derive requirements.
Data Exposure Surface (SAFE): While the skill uses external network tools (Perplexity, Context7, Firecrawl) and has file system access (Read, Glob, Grep), these are consistent with its primary purpose of research and no malicious data exfiltration patterns were detected.

domain-research