The past 48 hours have seen an intense focus on stability across OpenClaw's most heavily-used integrations. Over a dozen fixes address edge cases in WhatsApp message handling, Android phone features, and core gateway reliability — the kind of systematic polish that transforms experimental software into daily-driver infrastructure.
OpenClaw has grown from a hobbyist project to infrastructure serving hundreds of thousands of users across messaging platforms. At this scale, edge cases become daily occurrences — Unicode boundary issues in Telegram, timestamp precision in WhatsApp, race conditions during QR pairing. Each fix addresses real user pain points reported through GitHub issues.
Three separate fixes address WhatsApp edge cases:
843e3c1 — Restores append recency filter lost during the extensions refactor, plus proper handling of Long timestamps (JavaScript's number precision limits)b2e9221 — Fixes stale append inbox expectation in tests, ensuring CI catches real regressions9d3e653 — Handles 515 Stream Error during QR pairing, a frustrating first-run experience issuePR #47274 by Ayaan Zaidi preserves word boundaries when rechunking HTML for Telegram. When messages exceed Telegram's length limits and need splitting, naive chunking could break words or HTML tags. This fix ensures splits occur at natural boundaries.
PR #44073 adds calllog.search support for Android nodes — enabling AI assistants to query call history with filters. This continues the pattern of making smartphone features accessible to agents: camera, location, notifications, and now call logs.
a472f98 — Hardens remote CDP (Chrome DevTools Protocol) probes for browser automation26e0a3e — Skips Control UI pairing when auth.mode=none, fixing a startup issue for headless deployments5c5c64b — Deduplicates repeated tool call IDs for OpenAI-compatible APIs, preventing confusion when multiple calls have the same IDThese fixes represent the unsexy but essential work of production software engineering. No one tweets about timestamp precision handling, but it's the difference between "my AI assistant works reliably" and "it's flaky, I went back to doing things manually."
OpenClaw's issue tracker shows this pattern: users report specific scenarios ("WhatsApp messages arrive out of order when..."), contributors trace the root cause, and targeted fixes land. The project has enough adoption to surface edge cases and enough contributor momentum to address them.
Open PRs suggest continued focus on:
For users running OpenClaw in production: update to the latest release to pick up these stability improvements. For contributors: the issue tracker surfaces real user pain points waiting to be addressed.