openclaw-gitcode-pr-monitor

Monitor GitCode PRs for one or more repos, auto-run AI review via OpenClaw Gateway, post PR comments, and send notifications (DingTalk + WeCom).

This package looks like a legitimate GitCode PR monitor and reviewer, but the published metadata is incomplete. Before installing or running it: - Verify you have an official OpenClaw CLI binary and confirm the OPENCLAW_CMD used by scripts points to a trusted binary. The scripts call openclaw message send and openclaw agent which will transmit content externally. - Provide and protect the GitCode token: the scripts expect $HOME/.openclaw/workspace/data/gitcode-token.txt. Ensure this file is readable only by the intended user (chmod 600) and stored in a secure workspace. - Confirm you want automatic posting of comments and notifications. The submit-pr-comment.sh will post the full generated report as a PR comment using your GitCode token. - Ensure required helper binaries are present (curl, jq). The metadata doesn't declare them. - Inspect ~/.openclaw/openclaw.json and your OpenClaw messaging channel configurations so the TARGET_WECOM/TARGET_DINGTALK bindings do what you expect and do not leak reports to unintended recipients. - Because the metadata omitted required credentials/config paths and required binaries, treat this package as 'suspicious' until you confirm and secure the above items. If you need higher assurance, request the publisher to update the skill metadata to declare required env vars, credentials, and binaries (GitCode token, openclaw CLI, notification targets) and to document expected network endpoints.

Type: OpenClaw Skill Name: openclaw-gitcode-pr-monitor Version: 0.1.2 The skill bundle provides a legitimate pipeline for monitoring GitCode Pull Requests and performing automated AI-driven code reviews. It consists of shell scripts (`monitor-gitcode-pr.sh`, `code-review-robust.sh`, `submit-pr-comment.sh`) that poll the GitCode API, trigger the OpenClaw agent to analyze code diffs, and post the resulting reports back to the PR and notification channels like DingTalk and WeCom. The scripts handle authentication via a local token file and implement standard operational features like session isolation and file-based locking. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found.