product-interview
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads untrusted data from the local repository to inform product decisions, which could allow malicious instructions in those files to influence agent behavior.\n
- Ingestion points: The skill explores
CLAUDE.md, project design documents, existing UX components, and previously shipped specifications.\n - Boundary markers: Absent; the skill does not wrap external content in delimiters or include instructions to ignore embedded prompts during analysis.\n
- Capability inventory: The skill can write files to
meta/specs/, executegitcommands, and invoke other functional skills.\n - Sanitization: Absent; no validation or filtering is performed on the content read from project files.\n- [COMMAND_EXECUTION]: The skill executes shell commands to manage version control for generated artifacts.\n
- Evidence: Step 6 instructs the agent to run
git addandgit committo store the final specification file.
Audit Metadata