init
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes internal Node.js scripts (
install.mjsanddoctor.mjs) located within the skill's package directory to manage agent installation, validation, and removal processes. - [INDIRECT_PROMPT_INJECTION]: The skill surveys the target repository to identify build, test, lint, and typecheck commands. This creates a vulnerability surface where a malicious repository could provide harmful commands to be executed by the agent.
- Ingestion points: Repository survey in Step 3 which searches for project-specific commands and conventions.
- Boundary markers: The instructions define specific goals for the survey and use a placeholder
PACKAGEto scope script paths, though no strict data delimiters are applied to the repository content itself. - Capability inventory: The skill has shell execution capabilities, including running discovered check commands and internal setup scripts.
- Sanitization: The instructions explicitly direct the agent to skip commands requiring network services, credentials, live servers, or writes beyond build output.
- [DYNAMIC_EXECUTION]: In Step 5, the skill performs dynamic execution by running commands discovered at runtime from the repository configuration. Although it attempts to limit execution to "safe check commands" with a time limit, the actual content of these commands is determined by the external repository data.
Audit Metadata