OpenClaw 2026.4.12: Active Memory, LM Studio Support, Codex Provider & More

OpenClaw 2026.4.12 is a massive update packed with new features that fundamentally expand what your agent can do. This is one of the most important releases of the month.

Highlights

🧠 Active Memory Plugin

The new Active Memory plugin gives OpenClaw a dedicated memory sub-agent that runs automatically before each reply. Your agent now proactively pulls in relevant preferences, context, and past details — without you having to say “remember this” or “search memory.”

Features include configurable message/recent/full context modes, live /verbose inspection, and opt-in transcript persistence for debugging.

🖥️ LM Studio Provider

Local models get first-class support with the bundled LM Studio provider. Includes onboarding, runtime model discovery, stream preload support, and memory-search embeddings for self-hosted OpenAI-compatible models.

🤖 Codex Provider

The bundled Codex provider brings Codex-managed auth, native threads, model discovery, and compaction. Use codex/gpt-* models natively while openai/gpt-* stays on the normal OpenAI path.

🍎 macOS Talk: Local MLX Speech

macOS users get an experimental local MLX speech provider for Talk Mode with local utterance playback and interruption handling.

🔒 Security Fixes

• Removed busybox/toybox from interpreter-like safe bins
• Prevented empty approver lists from granting authorization
• Broadened shell-wrapper detection and blocked env-argv injection

Other Improvements

• Convex-backed pooled Telegram credential leasing for QA
• CLI exec policy management commands
• Matrix partial streaming with live typewriter animation
• Docs i18n improvements
• Dreaming improvements — consume heartbeat events exactly once, stop re-ingesting own transcripts

How to Update

npm install -g openclaw@latest
openclaw gateway restart