The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses gh (GitHub CLI) and git to perform diagnostic tasks. It retrieves workflow details, failure logs, and commit history to identify breaking changes. These actions are within the scope of the skill's described purpose.
[DATA_EXFILTRATION]: No data exfiltration patterns were identified. The network operations are limited to the GitHub API via the official CLI tool, and no sensitive local files are accessed or transmitted.
[PROMPT_INJECTION]: The skill processes untrusted external data, such as workflow logs and PR descriptions, which presents a surface for indirect prompt injection. However, given the skill's primary purpose as a log analyzer, this is an inherent risk that does not suggest malicious intent in the skill itself.

debug-gha-analysis