← Back to Blogs
GH PRs

OpenClaw Development Digest: Reliability, Performance, and New Features Across Recent Merges

18:30–00:30 UTC May 2, 2026

OpenClaw Development Digest: Reliability, Performance, and New Features Across Recent Merges

The openclaw/openclaw repository recently saw a flurry of activity, with numerous Pull Requests merged that collectively enhance the platform's reliability, performance, and feature set. This digest provides a detailed overview of these changes, focusing on how they address user needs, resolve pain points, and push the OpenClaw ecosystem forward.

From critical bug fixes ensuring accurate system status reporting to new capabilities like server-side voice dictation and expanded messaging channel support, these updates reflect a commitment to a more robust and user-friendly agent development and deployment experience.

Merged PRs

  • [AI-assisted] fix(feishu): probe status with account credentials (Original PR)
  • fix(pr): prefer verified GitHub commits for prep pushes (Original PR)
  • fix: Windows stale-import guard still generates an invalid PowerShell regex, so the intended... (Original PR)
  • fix(cron): persist manual run ids in history (Original PR)
  • fix(slack): ignore no_reaction in remove helpers (Original PR)
  • fix(queue): reset turns bypass steer/followup queue when active run exists (Original PR)
  • fix(gateway): preserve every client tool call when agent calls multiple tools per turn (#52288) (Original PR)
  • fix: reduce WebUI session latency churn (Original PR)
  • fix: keep media provider inventory internal (Original PR)
  • Show workspace bootstrap context in Codex prompt snapshots (Original PR)
  • feat(webchat): add server-side dictation (Original PR)
  • Fix Trinity main-session compatibility mismatch (Original PR)
  • fix(channels): clarify remote install hints (Original PR)
  • fix(gateway): keep requested plugin tools invokable (Original PR)
  • Rename Codex runtime prompt snapshot directory (Original PR)
  • fix(active-memory): fast-fail stalled recall paths (Original PR)
  • Improve Codex happy path prompt snapshots (Original PR)
  • Wizard: bound hatch TUI timeout (Original PR)
  • fix(gateway): memoize plugin descriptor config keys (Original PR)
  • fix(heartbeat): make phase scheduling active-hours-aware (#75487) (Original PR)
  • feat(whatsapp): support newsletter targets in message tool (Original PR)
  • fix(gateway): reuse subagent registry snapshot in session listing (Original PR)
  • fix(gateway): surface unreachable status diagnostics (Original PR)
  • ci: fold parity into QA release validation (Original PR)
  • test: exercise Clownfish any-PR automerge opt-in (Original PR)
  • Add agent visibility to skills check (Original PR)
  • [codex] Fix Codex OAuth status auth label (Original PR)

Key Changes

This batch of merged PRs introduces a mix of new features, crucial bug fixes, and performance enhancements:

New Features and Capabilities

  • Server-Side Dictation for WebChat: A significant addition is the chat.transcribeAudio Gateway RPC, enabling server-side microphone dictation in the Control UI/WebChat. This leverages OpenClaw's existing audio transcription pipeline, offering a privacy-friendly and consistent voice input experience across browsers, addressing limitations of native browser APIs. ([PR #76021], [Issue #47311])
  • WhatsApp Newsletter Support: The message tool for WhatsApp now supports sending messages to Newsletter/Channel targets (e.g., @newsletter JIDs). This expands the utility of OpenClaw agents for broadcast communication, fulfilling a long-standing feature request. ([PR #73393], [Issue #13417])
  • Enhanced Prompt Snapshots: Improvements to Codex runtime prompt snapshots now include OpenClaw's workspace bootstrap context (like SOUL.md, TOOLS.md) and the model-bound layer stack, making prompt audits more comprehensive for developers. ([PR #76302], [PR #76229])
  • Agent-Specific Skills Check: The openclaw skills check --agent <id> command provides granular visibility into skill readiness for specific agent runtimes, aligning skills list and skills info with agent allowlists. This empowers operators with better control and diagnostics for agent capabilities. ([PR #75983])
  • Gateway Unreachable Diagnostics: The status command now provides a no_gateway_reachable warning with actionable diagnostics when the Gateway is unreachable, guiding users to troubleshoot more effectively. ([PR #74691])

Reliability and Stability Fixes

  • Feishu Health Check Accuracy: Corrected an issue where Feishu health checks incorrectly reported

References

Pull Requests