openclaw-self-healing and openclaw-agent-doctor
These are complementary tools: the self-healing system provides automated recovery mechanisms while the diagnostic toolkit enables proactive problem identification and manual remediation, creating a layered resilience approach for OpenClaw agents.
About openclaw-self-healing
Ramsbaby/openclaw-self-healing
AI-powered self-healing system for OpenClaw Gateway • 4-tier autonomous recovery • macOS & Linux
Implements LLM-agnostic root-cause analysis (Claude, GPT-4, Gemini, Ollama) that diagnoses and auto-fixes corrupted configs, rate limits, and dependency failures — preventing crash loops before escalation. The 5-tier architecture progresses from instant restarts through HTTP health checks to PTY-based log analysis and optional multi-channel alerts (Discord/Slack/Telegram), with Prometheus metrics for observability across macOS, Linux, and Docker deployments.
About openclaw-agent-doctor
AlekseiUL/openclaw-agent-doctor
🏥 Self-diagnostic toolkit for OpenClaw AI agents. Find and fix problems before they find you.
Implements comprehensive system health checks across 7 categories (memory, crons, config, files, gateway, system, security) with a curated database of 28+ common problems and their concrete solutions. Operates as an OpenClaw skill that performs read-only diagnostics with user-confirmed fixes, supporting risk assessment for each remediation, and includes a standalone bash script option for CI/CD integration without requiring agent interaction.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work