co-build
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The procedure explicitly instructs the agent to perform file edits and run tests. These capabilities are required for the skill's purpose but involve direct system interaction and code execution.
- [PROMPT_INJECTION]: The skill processes task descriptions from
CHANGE.mdand specifications fromSPEC.md, creating a surface for indirect prompt injection. - Ingestion points: Task definitions in
CHANGE.md(§T) and documentation inSPEC.md(§V, §I, §T). - Boundary markers: None identified in the instructions; the agent is not directed to use specific delimiters or ignore embedded commands.
- Capability inventory: File system modification and test execution commands (SKILL.md).
- Sanitization: None described; the agent is expected to use the raw task text for planning and execution.
Audit Metadata