fix-linter-test
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the Bash tool for building and running Docker containers. This includes commands like 'docker buildx build' and 'docker run' which use local Dockerfiles and mount the project root as a volume to reproduce linter errors.
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing data from files within the repository. 1. Ingestion points: Test files, descriptor YAMLs, fixtures, and custom linter classes located in 'megalinter/' and '.automation/' directories. 2. Boundary markers: The instructions do not specify any delimiters or safety warnings to ignore embedded instructions within the files being read. 3. Capability inventory: The agent possesses Read, Grep, Glob, Edit, and Bash tools, giving it the ability to modify the environment and execute code based on the data it processes. 4. Sanitization: There is no requirement or logic for sanitizing or validating the content of the project files before the agent analyzes and acts upon them.
Audit Metadata