customer-research
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to ingest and synthesize data from untrusted third-party sources, such as Reddit posts, YouTube comments, X (Twitter) threads, and community review sites. Maliciously crafted content on these platforms could contain instructions designed to hijack the agent's behavior during synthesis.
- Ingestion points: Untrusted data is brought into the agent's context through tools like reddit_scrape, youtube_comments_search, x_tweets_search, and web_pages_scrape as detailed in SKILL.md and references/source-playbooks.md.
- Boundary markers: The skill does not prescribe the use of delimiters or 'ignore' instructions for the ingested data; conversely, it explicitly mandates the capture of 'verbatim language' in its critical rules.
- Capability inventory: The skill enables searching Google, scraping arbitrary URLs, and synthesizing findings into complex reports and personas.
- Sanitization: There are no instructions for sanitizing, filtering, or escaping the external content before it is processed by the model.
- [EXTERNAL_DOWNLOADS]: The skill fetches content from well-known public platforms and social media services (Reddit, YouTube, Twitter, G2, etc.) to gather research data.
Audit Metadata