The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs shell-level operations including git diff and git status to establish a baseline of changes. It also identifies and executes local tests, build scripts, and application logic as part of its verification process.
[EXTERNAL_DOWNLOADS]: For its 'suite mode', the skill instructs the agent to clone the fable-method repository from https://github.com/Sahir619/fable-method if the necessary evaluation tools are not already present in the environment.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it is designed to process untrusted data (external reports and arbitrary code diffs) and use that data to drive high-privilege actions like code execution.
Ingestion points: The skill reads conversation history, pasted reports from other agents, and file system content (diffs).
Boundary markers: No explicit markers or warnings are used to prevent the agent from following instructions embedded within the 'work' it is judging.
Capability inventory: The skill can read files, execute git commands, and run arbitrary shell commands defined in the project's tests or build configuration.
Sanitization: There is no evidence of input validation or sanitization of the content being verified.

fable-judge