The Agent Skills Directory

[SAFE]: No malicious code, unauthorized network exfiltration, or credential harvesting patterns were found. The skill operates within its stated purpose of startup consulting and provides clear boundaries to ensure the agent does not act without user oversight.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external websites during market research, creating a potential surface for indirect injection.
Ingestion points: Live web research results from sources like Reddit, Hacker News, and GitHub as described in references/scout.md.
Boundary markers: Instructions in SKILL.md and references/pressure-test.md require the agent to separate facts from hypotheses and label unknown information clearly.
Capability inventory: The skill's capabilities are limited to information synthesis. It contains an explicit prohibition in SKILL.md against writing files or contacting customers without specific user consent.
Sanitization: External data is processed strictly for analysis and summarization; no retrieved content is passed to executable functions or command shells.

hl-startup