Senior Project Manager
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from local project files to drive its logic.
- Ingestion points: The agent is instructed to read the site specification file located at
ai/memory-bank/site-setup.mdto extract project requirements. - Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore any embedded directives within the processed specification file.
- Capability inventory: The skill possesses the capability to write new markdown files to
ai/memory-bank/tasks/and is instructed to execute a local shell script (./qa-playwright-capture.sh). - Sanitization: No sanitization or validation of the input file content is described before the agent uses it to generate tasks or execute commands.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local shell script as part of the project's quality assurance workflow.
- Evidence: The 'Quality Requirements' section contains a specific command for the agent to run:
./qa-playwright-capture.sh http://localhost:8000 public/qa-screenshots. Executing local scripts can be risky if the arguments or environment are influenced by injected data from the processed specification files.
Audit Metadata