competitive-intelligence
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a workflow that involves processing external, potentially untrusted data such as competitor evidence, customer alternatives, and market signals.\n
- Ingestion points: The skill ingests external data during the 'evidence' workflow phase and via the 'Inputs' section in
references/decision-model.md, which includes competitor and substitute evidence.\n - Boundary markers: There are no explicit delimiters or 'ignore embedded instructions' markers for external data; however, the
SKILL.mdfile mandates an 'Evidence discipline' to classify claims as fact, inference, or assumption.\n - Capability inventory: The skill references execution of internal scripts such as
../../scripts/dynamic_router.pyand../../scripts/neural_router.pyfor cross-functional routing and specialist selection.\n - Sanitization: The skill does not define specific sanitization routines for external content, relying instead on the agent's ability to classify and separate evidence from inference as a manual check.
Audit Metadata