OpenClaw v2026.4.1 landed on April 1, 2026. This release focuses on making agents easier to manage in real time, adding private web search out of the box, and bringing enterprise-grade guardrails to the open-source agent framework.
The headline feature is /tasks — a built-in task board that lives inside your chat session. Before this release, running background jobs meant switching to a separate terminal or dashboard. Now you type /tasks and see every queued, running, and completed job in one place.
This matters for teams running multi-agent workflows. You can kick off a research agent, a writing agent, and a code review agent in parallel, then check their progress without leaving the conversation. Each task shows its status, duration, and output summary.
Web search is one of the most requested agent capabilities. v2026.4.1 ships with a bundled SearXNG provider plugin that gives your agents web search without relying on commercial APIs.
SearXNG aggregates results from multiple search engines while keeping queries private. No API keys needed — just enable the plugin and your agents can search the web. This is especially useful for self-hosted deployments where data privacy is a hard requirement.
Enterprise users running agents through AWS now get native support for Amazon Bedrock Guardrails. You can define content filters, denied topics, and word filters that apply at the provider level — meaning every agent response goes through your guardrail configuration before reaching the user.
This is a significant step for regulated industries. Healthcare, finance, and government teams can now use OpenClaw agents with the same content safety controls they use for their other Bedrock workloads.
Talk Mode gets a quality-of-life upgrade: Voice Wake. On macOS, you can now activate your agent with a wake word instead of clicking a button or pressing a hotkey. It turns OpenClaw into a hands-free assistant that listens for your command.
Voice Wake uses macOS native speech recognition, so processing happens on-device. No audio is sent to external servers until you actually trigger the agent. This keeps the always-listening part completely local.
Long conversations with agents can eat memory and slow down responses. This release adds gateway.webchat.chatHistoryMaxChars — a configuration option that lets you set a hard limit on how much chat history is sent to the model.
Set it to 50,000 characters for detailed context, or 10,000 for faster responses. The gateway automatically truncates older messages while keeping recent ones intact. This is particularly useful for agents that run 24/7 and accumulate long conversation histories.
v2026.4.1 also fixes critical task registry maintenance issues that could cause background tasks to hang or report incorrect status. If you were experiencing stale tasks in previous versions, this update resolves it.
If you are running OpenClaw via Docker, pull the latest image. For manual installations, run your standard update command. The SearXNG plugin and Voice Wake feature are opt-in — enable them in your configuration after updating.
Full changelog is available on the GitHub release page.
The /tasks command opens a chat-native background task board inside your session. You can queue long-running jobs, check their status, and get notified when they finish — all without leaving the chat interface.
SearXNG is a privacy-respecting metasearch engine. OpenClaw now bundles it as a provider plugin, so your agents can search the web without sending queries to Google or Bing directly. It aggregates results from multiple search engines while keeping your searches private.
Yes. Amazon Bedrock Guardrails require an AWS account with Bedrock access configured. Once set up, you can define content filters and topic blocks that apply to any model your OpenClaw agents use through Bedrock.
Open OpenClaw settings, navigate to Talk Mode, and enable the Voice Wake option. Once active, you can trigger your agent with a wake word instead of pressing a button. This feature is macOS-only in this release.
Skip the setup. Pick a template and deploy in 60 seconds.
Pick a role. Your AI employee starts working in 60 seconds. WhatsApp, Telegram, Slack & Discord. No setup required.
Get Your AI Employee