knock-product-messaging-strategy
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze external product data (codebase, event schemas, and product state) to generate messaging recommendations in
knock-plan.md. This represents a potential surface for indirect prompt injection, as malicious content or metadata embedded within the analyzed product assets could attempt to influence the agent's strategic recommendations or resource definitions. - Ingestion points: The skill reads product codebases, events, roles, and lifecycle data as defined in
SKILL.mdandrules/data-foundation-and-triggers.md. - Boundary markers: The instructions do not explicitly define boundary markers or 'ignore' directives for untrusted code/event data during the discovery phase.
- Capability inventory: The skill can create or update resources in Knock using MCP tools or the CLI, and it writes detailed plans to the local file system (
knock-plan.md). - Sanitization: There are no explicit instructions for sanitizing or escaping content extracted from the product codebase before it is used to draft workflow templates or guides.
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to use Knock MCP tools or the Knock CLI to build resources. These are legitimate, intended functionalities for a vendor-provided skill and are only triggered upon explicit user request.
Audit Metadata