company-tech-profile
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including web search engines (Exa, Tavily, Brave, Jina) and academic/patent databases (SKILL.md, references/source-routing.md). The instructions provide boundary guidance by requiring the agent to separate verified facts from inferences and to corroborate marketing signals (SKILL.md, references/quality-gates.md). The agent has the capability to write files to local artifacts directories (references/workflow.md). No programmatic sanitization is specified beyond logic-based triangulation of evidence (references/domain-playbooks.md).
- [COMMAND_EXECUTION]: The instructions suggest the optional use of external utilities like pandoc for document format conversion (references/deliverables.md).
- [EXTERNAL_DOWNLOADS]: The skill references and utilizes well-known external research and search services to gather evidence for the R&D assessment (SKILL.md, references/source-routing.md).
Audit Metadata