OpenClaw v2026.6.8 Released: 373 Commits of Richer Channels, Safer Routing, and Reliable Agents
By AgentRiot Editorial
OpenClaw ships v2026.6.8 with 185 merged PRs and 373 commits. Highlights include richer Telegram and WhatsApp delivery, safer model routing with GLM-5.2 and Claude Haiku 4.5, more reliable agent execution, native usage footers, and improved memory resilience.
OpenClaw v2026.6.8 shipped with 373 commits and 185 merged pull requests. The release is a reliability and polish drop: Telegram and WhatsApp messaging got richer, the model catalog added GLM-5.2 and Claude Haiku 4.5, agent execution became more predictable, and a long list of edge-case crashes and silent failures were fixed.
What changed
Telegram and WhatsApp now handle rich text
Telegram messages in OpenClaw were historically plain text. Tables, lists, and blockquotes either rendered as garbled markdown or were stripped entirely. v2026.6.8 preserves intentional line breaks, renders tables and lists natively, and supports expandable blockquotes. WhatsApp ACP bindings, which previously ignored configured transports, now honor the user's setup. Thread creation via CLI remapping and streaming preview edits also stopped crashing mid-send.
Agent runs no longer die on edge cases
A set of fixes keeps agent execution on the correct recovery path. Account-scoped DM sends now respect the right channel policy. Generated media completions, auto-reply message-tool final replies, reset archive fallback reads, restart shutdown aborts, yielded subagent pauses, and session identity prompts all behave more predictably. The common thread: previously these paths would either hang silently or drop the user into an unrecoverable state.
Safer model routing
The catalog now includes GLM-5.2 and Claude Haiku 4.5. Provider IDs are normalized across OpenRouter and Google Vertex paths. Managed SecretRef authentication was introduced, model browsing is now bounded to prevent runaway catalog loading, and OpenAI/Anthropic tool-schema recovery handles malformed or post-hook schemas without crashing the provider.
Usage footers that actually work
The /usage command and reply payload hooks now ship a native full footer renderer with a default template, fixed-decimal formatting, and credential-aware limits. Broken templates surface warnings instead of producing silent bad output.
Web search defaults you can predict
Parallel Free, DuckDuckGo, Ollama, and Codex Hosted Search are now explicit opt-ins. Previously, if no API-backed provider was configured, OpenClaw would silently fall back to one of these key-free sources, which surprised users who expected a configured provider to be used.
UI and mobile cleanup
Workspace files start collapsed. WebChat backscroll survives streaming without jumping. The desktop session picker stays interactive above the workbench. Reset arguments survive dispatch. On iOS, stale foreground Gateways reconnect automatically.
Memory and state resilience
Oversized OpenAI embedding batches split before hitting HTTP 431 errors. QMD search stays available in transient mode. SQLite avoids WAL mode on NFS volumes, preventing corruption on network-mounted state directories. Full reindexes preserve rollback and cache recovery paths.
Other fixes worth noting
- Discord auto-thread titles get a 60-second timeout and a 4,096-token reasoning-model output budget, clamped to the selected model cap.
- Gemini CLI uses the selected OpenClaw OAuth/API-key auth profile in an isolated runtime home, preventing ambient Google machine credentials from overriding the chosen profile.
- Slack outbound
message_senthooks now fire correctly, mirroring Telegram behavior. - Feishu dynamic-agent routes re-resolve after persisted binding reuse.
- Managed plugin installs and updates repair missing required platform packages, including Codex platform binaries.
- Hono updated to 4.12.25.
- False downgrade prompts when the latest tag cannot resolve are now avoided.
Release metadata
| Field | Value |
|---|---|
| Version | v2026.6.8 |
| Published | June 16, 2026 at 16:32 UTC |
| Commits since v2026.6.6 | 373 |
| Merged PRs | 185 |
| GitHub stars | 379,151 |
| npm | [email protected] |
| Registry | https://registry.npmjs.org/openclaw/-/openclaw-2026.6.8.tgz |
| Integrity | sha512-iziR8fi69+ojrtX7FYYvTpkGcVnmyLpIhvchgt5LFkkdHVWw973XAAekKVZ3/xQJ5FG4NwgHkXL0LLTrgsNOSQ== |
How to update
openclaw update
For Docker-based deployments, pull the latest image and restart your container. The full release validation report and CI evidence are available at the OpenClaw releases repository.
Bottom line
v2026.6.8 is a reliability and polish release. The headline wins are better Telegram and WhatsApp delivery, safer model routing with new catalog entries, and a more predictable agent runtime. If you run OpenClaw in production or rely on channel integrations, this is a recommended upgrade.