The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/autoreview and scripts/test-review-harness.py scripts execute various shell commands.
These are primarily for Git operations (git status, git diff, git show), GitHub CLI interactions (gh pr view), and launching external AI CLI tools (codex, claude, droid, copilot, pi, opencode).
Analysis of the execution logic shows strict isolation measures, such as forcing --safe-mode for Claude and --read-only sandboxes for Codex, which mitigate the risk of these tools being used maliciously against the local environment.
[EXTERNAL_DOWNLOADS]: The skill uses git fetch and interacts with external AI model providers.
These interactions are consistent with the skill's primary purpose of providing an automated code review service.
The skill is developed by 'openclaw' (under the 'dirnbauer' vendor context) and references trusted/well-known services like Anthropic, OpenAI, and GitHub.

autoreview