atr-analyze
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill's fundamental purpose is to execute arbitrary commands supplied to the
--cmdparameter. This provides a functional path for the agent to run any system command, such as build scripts, test suites, or linters. - [INDIRECT_PROMPT_INJECTION]: The skill specifies that the agent analyzes the complete output of the commands it runs to generate summaries and recommendations. This exposes the agent to indirect prompt injection if the output of a command (e.g., log files or test results) contains adversarial instructions.
- Ingestion points: The full, raw output of the executed commands (SKILL.md).
- Boundary markers: None are specified in the documentation to distinguish between command output and analysis instructions.
- Capability inventory: The AI agent has file system read access, as it is instructed to 'Read relevant source files when failures occur' (SKILL.md).
- Sanitization: There is no documentation of sanitization or filtering being applied to the command output before the AI processes it.
Audit Metadata