pm-mvp
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads external markdown documents (such as '优先级排序报告.md' and '需求调研报告.md') to extract product requirements.
- Ingestion points: Step 1 reads files from the
docs/01-需求调研/directory. - Boundary markers: None are present to delimit the external content or instruct the agent to ignore embedded commands.
- Capability inventory: The skill has
Writeaccess to the file system andBashexecution capabilities. - Sanitization: No specific sanitization or validation of the input file content is performed. While this creates a theoretical attack surface, the risk is inherent to tools processing user-provided documentation.
- [COMMAND_EXECUTION]: The skill includes a Bash script in the preamble that checks for the existence of a local file ('docs/01-需求调研/优先级排序报告.md'). This is a benign environment check used to verify that prerequisite documentation is available before proceeding with the MVP planning workflow.
- [DATA_EXFILTRATION]: The skill's file operations are restricted to specific paths within the project's documentation directory. There are no network operations, and no access to sensitive system paths (such as SSH keys or environment variables) was detected.
Audit Metadata