· Updated
/fast toggle enables session-level fast-mode switching: on Claude it maps to Anthropic's service_tier; on OpenAI and Codex it shapes requests to their fast tier.Ollama, vLLM, and SGLang move to the provider-plugin architecture, removing local model wiring from the core so each provider owns its onboarding, discovery, and model-picker hooks — shrinking the shared footprint. The release also patches two CVEs: GHSA-99qw-6mr3-36qr (implicit workspace plugin auto-load disabled) and GHSA-pcqg-f7rg-xfvv (invisible Unicode escapes in approval prompts), and switches device pairing to ephemeral tokens.
Update to v2026.3.12 and open the new dashboard — the command palette and modular session views make it a practical daily surface for multi-channel agent deployments.
OpenClaw 2026.3.12 🦞 🎛️ dashboard v2 — slick new control UI ⚡ /fast mode for models 🔌 ollama/sglang/vllm → plugins (core goes on a diet) 🛡️ device tokens now ephemeral ⏰ cron + windows reliability fixes touch grass? nah, touch main 🌿https://t.co/OCczevtDhe
More like this
