kai-analytics
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read various files from the project root (MARKETING.md, CLAUDE.md, README.md, PROJECT.md, package.json) and "any project files" to understand the product context. \n
- Ingestion points: SKILL.md (Phase 0 and Phase 1) specifies reading local project files to gather context. \n
- Boundary markers: The instructions lack boundary markers or warnings to ignore instructions that might be embedded within the project files being read. \n
- Capability inventory: The skill primarily generates documentation (tracking plans, UTM guides, dashboard specs) and does not perform shell execution or network operations. \n
- Sanitization: There is no mention of sanitizing or validating the content of the files read from the codebase before they are processed by the agent.
Audit Metadata