OpenClaw Elite Watcher

Zero-latency intelligence engine for the OpenClaw ecosystem. Monitors core protocol commits from Peter Steinberger and top developers, distilling raw code di...

This skill appears misleading: it advertises real-time monitoring but its code only creates a static report and does not query GitHub or X. Before installing or running it, consider: 1) Inspect and, if needed, change REPORT_PATH to a safe, configurable directory (or run in a sandbox/container). 2) Verify whether the skill will ever fetch network data—look for later versions that add requests usage; require review before granting network access. 3) Expect it to create files under the specified path; back up or audit that directory. 4) Ask the author (or reject) for a clear list of required credentials and exact network endpoints if you want true repo/X monitoring. 5) If you still want to test, run it in an isolated environment (container or VM) and monitor outbound network connections and file writes. Because the implementation is inconsistent with its description, do not grant broad trust or production access without a code review and clarification from the publisher.

Type: OpenClaw Skill Name: openclaw-elite-watcher Version: 1.0.0 The `watcher.py` script is suspicious because it includes a prompt injection attempt against the OpenClaw agent within the generated `SKILL.md` report. Specifically, it instructs the agent to 'discover new tools' and 'complete preliminary integration' of `agent-twitter-client` (elizaOS), attempting to manipulate the agent's future actions and tool usage beyond the skill's stated purpose. Furthermore, the script misleadingly claims to monitor GitHub and X.com but only generates a static report without performing any actual network requests or monitoring, and it uses a hardcoded, user-specific path (`/Users/asdc163/.openclaw/workspace/intel_reports`) for report generation, indicating poor practice and potential portability issues.