Add OpenClaw community tools section — self-healing, memory, and self-evolving for Claude Code agents

[Ramsbaby]

What this adds

A new ### 🔧 Community Tools & Extensions subsection under 🛠️ Claude Code & Model Context Protocol (MCP) featuring three OpenClaw repos for Claude Code agent infrastructure:

• openclaw-self-healing (32 ⭐) — Self-healing infrastructure for Claude Code AI agents. Auto-recovers from failures using a 4-tier escalation system: watchdog, health checks, AI diagnosis, and alerts.
• openclaw-memorybox (8 ⭐) — Persistent memory & skill storage for Claude Code. Saves context across sessions via slash commands.
• openclaw-self-evolving (2 ⭐) — AI agent that analyzes its own codebase and generates weekly self-improvement proposals as GitHub Issues.

Why this fits here

These are infrastructure-level tools that make Claude Code agents more reliable, persistent, and autonomous — complementing the official Claude Code resources already listed. All are open-source (MIT) and publicly available on GitHub.

The new subsection is placed after the MCP entries since these tools relate to extending Claude Code's runtime capabilities rather than model access.