go-to-market
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from upstream files including
messaging,business-model, andcustomer-blueprint. While this presents an attack surface where instructions embedded in those files could influence behavior, the skill includes hard rules requiring all output to be cited from source evidence, which significantly mitigates this risk. - Ingestion points: Reads content from
messaging,business-model, andcustomer-blueprintfiles within the workspace. - Boundary markers: Uses a structured rubric and citation requirements; however, it lacks explicit technical delimiters for external content.
- Capability inventory: Performs file writes (
GTM.md,go-to-market.json), generates a static site and MCP tools, and manages database-like tables (leads,campaigns). - Sanitization: Relies on the agent to strictly follow citation rules rather than restating or amending content, but no programmatic sanitization of input data is defined.
- [SAFE]: The skill operates entirely within a local workspace context and does not exhibit patterns of data exfiltration, obfuscation, or unauthorized network access. The generation of documentation and structural artifacts like MCP servers is consistent with the stated purpose of the 'Business OS' protocol.
Audit Metadata