startup-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a vulnerability surface by processing untrusted external data (web search results and company websites) to perform its analysis.
- Ingestion points: The skill instructions in SKILL.md (Step 1) prompt the agent to gather data from the web, company sites, and external platforms like LinkedIn or Crunchbase.
- Capability inventory: The agent leverages web search and page reading tools to retrieve this data.
- Sanitization: The instructions lack specific delimiters or "ignore embedded instructions" directives to mitigate risks from malicious content on external sites.
- Boundary markers: No explicit boundaries are defined for content retrieved from external sources.
- [EXTERNAL_DOWNLOADS]: The skill provides installation commands for the user to add the skill and its parent plugin from the vendor's repository.
- Installation source: himself65/finance-skills via npx.
- Execution method: User-initiated installation using standard Node.js package execution (npx).
Audit Metadata