2-minute diagnostic for issues you didn't know about—
dead skills, broken cron jobs, AGENTS.md burning token
One-time payment · Free retest after fixes · Credit card accepted
You think 30 skills installed, all working
Reality 18 are broken (missing deps/config) and burn token every chat
You think cron runs every morning at 8am
Reality timezone is UTC, it never triggered at the right time
You think AGENTS.md is well-written
Reality 14,000 chars = ~28K token per turn, extra $50+/month
AgentCoach scan complete OpenClaw 2026.3.8 / Gemini-2.5-Flash High priority (2) · 57 skills installed, 41 unusable (missing deps) · Memory poisoning risk: MEMORY.md contains security override keywords Medium (2) · Cron job daily-report wrong timezone (UTC vs Asia/Shanghai) · Sandbox off, exec tool unrestricted Suggestions (3) · AGENTS.md 14,755 chars → trim to save ~$30/mo token · MEMORY.md not updated for 6 days · No fallback model, agent crashes if primary fails ─────────────────────── Skills: 16/57 · Prompt: 41K chars · Tools: 37 · Cron: 4 Next: 1. Get fix steps 2. Set up weekly auto-scan
One command on your OpenClaw machine, 30 seconds to install.
AgentCoach collects config, uses your Agent for analysis. No sensitive data uploaded.
Each issue includes fix steps. Say "retest" after fixes—unlimited free reruns.
One-time, lifetime access
Requires OpenClaw (Linux / macOS)