senior-software-engineer
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands defined in the 'Verify:' field of tasks within 'specs/TASKS.md'.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted specification files which can influence its behavior.
- Ingestion points: 'specs/TASKS.md' and referenced files in the 'specs/' directory.
- Boundary markers: Absent; the skill is instructed to treat these files as 'immutable laws'.
- Capability inventory: Execution of arbitrary commands (Rule 6 'Verify:'), file modification (Rule 4), and git operations (Rule 7).
- Sanitization: Absent; the skill directly extracts and executes the verification string.
Audit Metadata