The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes arbitrary shell commands through the command-check verification method. These commands are defined in eval case files located within the .evals/cases/ directory and are executed during the regression testing phase.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting untrusted data from external sources to generate executable configurations.
Ingestion points: The skill reads 'gap report issue comments' and artifacts from learning-aggregator-ci to identify candidates for new evaluation cases in Create Mode.
Boundary markers: There are no specified delimiters or instructions to ignore embedded malicious prompts within the external data sources used for case creation.
Capability inventory: The skill has the capability to execute shell commands, read/write repository files, and interact with the GitHub API (PRs, issues, actions).
Sanitization: The skill lacks explicit sanitization or validation protocols for content extracted from issue comments before it is interpolated into new evaluation case files.
[EXTERNAL_DOWNLOADS]: The skill documentation guides users to install the gh-aw extension from GitHub's official repository, which is a well-known technology service.

eval-creator-ci