pre-commit-check
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingest staged files as data for review, creating a vector for instructions to be embedded in the code.
- Ingestion points: Staged repository files (SKILL.md).
- Boundary markers: Absent. There are no instructions to use delimiters or ignore instructions found within the reviewed code files.
- Capability inventory: Shell execution for builds, test suites, linters, and cloud operations such as Lambda invocation and CDK/SST deployment (SKILL.md).
- Sanitization: Absent. No filtering or validation of code content is performed before triggering commands.
- [COMMAND_EXECUTION]: The skill requires the agent to run multiple potentially dangerous shell commands.
- Evidence: The instructions direct the agent to 'Run the build', 'Run the linter', 'Run [tests]', 'Deploy to dev', and 'invoke the Lambda directly' (SKILL.md).
Audit Metadata