OpenClaw Changelog
Release notes, new features, and updates tracked daily. Sourced from the official OpenClaw changelog at openclaw.ai.
- App Updatevv2026.5.20
Add doctor warnings for hidden MCP server tools
The doctor utility now warns users when sandbox tool policies hide configured Model Context Protocol (MCP) server tools before provider requests are made.
Source ↗ - App Updatevv2026.5.20
Expose stale-running task maintenance details in CLI
The tasks maintenance CLI command now outputs detailed JSON explanations for retained and reconcile candidates, detailing backing-session, cron, CLI, and wedged-subagent states.
Source ↗ - Featurevv2026.5.20
Support per-agent lean local-model configuration
Added support for the experimental localModelLean configuration option on individual agents. This allows lean local-model mode to be enabled selectively rather than globally.
Source ↗ - Featurevv2026.5.20
Introduce bundled Policy plugin for workspace conformance
Added a bundled Policy plugin that enables policy-backed channel conformance checks, doctor lint findings, and opt-in workspace repair to streamline configuration management.
Source ↗ - Model Releasev2026.5.20-beta.1
Upgrade bundled Codex harness to version 0.132.0
Bumped the bundled Codex harness to @openai/codex 0.132.0 and refreshed the app-server model-list documentation to support the new catalog.
Source ↗ - Integrationv2026.5.20-beta.1
Add device-code OAuth login for xAI provider
Remote and headless setups can now authorize the xAI provider using device-code OAuth login, eliminating the need for a localhost browser callback.
Source ↗ - Integrationv2026.5.20-beta.1
Enable Discord voice sessions to follow configured users
Discord voice sessions can now follow configured users into voice channels with allowed-channel checks, multi-user handoff, bounded reconciliation, and DAVE recovery preservation.
Source ↗ - Integrationv2026.5.19
Add native DM draft previews for Telegram tool progress
Telegram integration now supports allowlisted native DM draft previews for transient tool progress while keeping final answers on the normal persistent delivery path.
Source ↗ - Featurev2026.5.19
Add CLI commands for building and validating tool plugins
Introduced defineToolPlugin alongside new CLI commands to build, validate, and initialize typed simple tool plugins with generated manifest metadata.
Source ↗ - Featurev2026.5.19
Enhance browser automation with modal dialog handling
The browser tool now surfaces pending and recently handled modal dialogs in snapshots, returns blockedByDialog status, and allows answering dialogs via CLI.
Source ↗ - Featurev2026.5.19
Introduce meme-maker skill with local and hosted rendering
Added a new meme-maker skill supporting curated template search, local SVG/PNG rendering, Imgflip hosted rendering, and Know Your Meme provenance links.
Source ↗ - App Updatev2026.5.19
Redesign Mac app Settings pages with card layouts
Redesigned the Mac app Settings pages with consistent card layouts, cached navigation, cleaner permissions, voice, skills, cron, exec, and debug panes.
Source ↗ - App Updatev2026.5.19
Upgrade Android Talk Mode to realtime voice sessions
Switched Android Talk Mode to use realtime Gateway relay voice sessions, featuring streaming mic input, realtime audio playback, tool-result bridging, and on-screen transcripts.
Source ↗ - Featurev2026.5.18
Expand QA-Lab with runtime parity scenarios
Added comprehensive runtime parity tiers and token-efficiency artifacts to the QA-Lab, including specific checks for Codex-vs-Pi compatibility and tool fixture coverage.
Source ↗ - App Updatev2026.5.18
Update Node.js requirement and Pi packages
Raised the minimum supported Node.js version to 22.19 and updated Pi packages to version 0.75.1 to ensure compatibility with the latest runtime features.
Source ↗ - App Updatev2026.5.18
Optimize Gateway startup and restart latency
Reduced restart ready latency by overlapping startup logging and plugin-service initialization with channel sidecars while maintaining strict readiness gating.
Source ↗ - Integrationv2026.5.18
Support HTTPS managed forward-proxy endpoints
The proxy system now supports HTTPS managed forward-proxy endpoints and introduces `proxy.tls.caFile` for configuring scoped CA trust for proxy TLS connections.
Source ↗ - App Updatev2026.5.18
Switch Android Talk Mode to realtime Gateway relay
Android's Talk Mode now utilizes realtime Gateway relay voice sessions, enabling streaming mic input, low-latency audio playback, and on-screen transcripts.
Source ↗ - Featurev2026.5.18
Add meme-maker and Python debugging skills
New built-in skills include a meme-maker with local SVG/PNG rendering and a Python debugging skill supporting pdb, breakpoint() inspection, and debugpy remote attachment.
Source ↗ - App Updatev2026.5.18
Redesign Mac app Settings and navigation
The Mac app Settings pages now feature consistent card layouts, cached navigation, and a persistent sidebar. Redesigned panes for permissions, voice, and debug improve spacing and usability.
Source ↗ - Featurev2026.5.18
Introduce typed tool plugins and CLI build tools
Added `defineToolPlugin` and new CLI commands (`build`, `validate`, `init`) to streamline the creation of typed tool plugins with generated manifest metadata and context factories.
Source ↗ - App Updatevv2026.4.27
OpenClaw adds Codex Computer Use and DeepInfra support (2026.4.27)
OpenClaw 2026.4.27 ships Codex Computer Use setup, DeepInfra as a bundled provider, Tencent Yuanbao and QQBot support, and manifest-first plugin metadata that reduces Gateway boot work.
Source ↗ - App Updatevv2026.4.26
OpenClaw ships browser realtime transport, Cerebras, asymmetric embeddings (2026.4.26)
OpenClaw 2026.4.26 ships a generic browser realtime transport contract with Google Live Talk sessions, Cerebras as a bundled provider, asymmetric embedding endpoint config, and provider-filtered model routing.
Source ↗ - App Updatevv2026.4.25
OpenClaw lands TTS upgrade, OpenTelemetry coverage, browser hardening (2026.4.25)
OpenClaw 2026.4.25 ships a TTS upgrade with /tts latest, chat-scoped auto-TTS, six new voice providers, OpenTelemetry coverage across the agent loop, browser automation hardening, and PWA / Web Push support.
Source ↗ - App Update
OpenClaw 2026.4.20 adds session pruning, cron fixes, subagent reporting
OpenClaw 2026.4.20 prevents OOM from cron backlog, handles cron modes correctly, adds failure context, and fixes PowerShell parsing on Windows for better stability.
Source ↗ - App Updatev2026.4.19-beta.1
Fix cross-agent subagent channel routing
Route subagent spawns through the target agent's bound channel account to prevent child sessions from incorrectly inheriting the caller's account in shared workspaces.
Source ↗ - App Updatev2026.4.19-beta.2
Isolate nested agent sessions
Scope nested agent work per target session to prevent long-running nested runs from blocking unrelated sessions across the gateway.
Source ↗ - App Updatev2026.4.19-beta.2
Improve OpenAI streaming usage reporting
Always send `stream_options.include_usage` on streaming requests to ensure local and custom OpenAI-compatible backends report accurate context usage instead of defaulting to 0%.
Source ↗ - Featurev2026.4.15
Add experimental local model lean mode
Introduce `agents.defaults.experimental.localModelLean: true` to disable heavyweight tools like browser and cron, reducing prompt size for weaker local models.
Source ↗ - Integrationv2026.4.15
Add GitHub Copilot embedding provider
Integrate a GitHub Copilot embedding provider for memory search, enabling plugins to reuse the transport while honoring remote overrides and token refresh.
Source ↗ - Featurev2026.4.15
Enable cloud storage for LanceDB memory
Add cloud storage support to `memory-lancedb`, allowing durable memory indexes to run on remote object storage instead of local disk.
Source ↗ - Featurev2026.4.15
Add Model Auth status dashboard
Introduce a Model Auth status card in the Control UI to monitor OAuth token health and provider rate-limit pressure at a glance.
Source ↗ - Featurev2026.4.15
Add Gemini text-to-speech support
Integrate Gemini TTS into the bundled Google plugin, supporting voice selection, WAV/PCM output, and automated setup guidance.
Source ↗ - Model Releasev2026.4.15
Add Claude Opus 4.7 support
Default Anthropic selections, Opus aliases, and CLI defaults to Claude Opus 4.7 for improved model performance.
Source ↗