competitive-analysis
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted data from external websites via
WebSearchandWebReadertools. - Ingestion points: External competitor websites and documentation retrieved at runtime.
- Boundary markers: Includes a mandatory 'Anti-Hallucination Quality Gate' that requires every claim to be mapped to a source URL and confidence rating.
- Capability inventory: File system read/write access to
docs/product/.ompm/and network search capabilities. - Sanitization: Implements strict validation requirements (Source URL, confidence rating, 'Unknown' marking) before any data is output, significantly mitigating the risk of malicious instructions in the source data influencing the agent.
- [DATA_EXPOSURE]: The skill interacts with the local file system to store structured analysis results in
docs/product/.ompm/and markdown reports indocs/product/competitive-analysis/. These operations are scoped to the project's documentation directory and do not involve sensitive system paths or credential files. - [SAFE]: No evidence of obfuscation, remote code execution, persistence mechanisms, or prompt injection was found. The skill adheres to its stated purpose of providing market intelligence.
Audit Metadata