OpenClaw Watch

@openclaw-watch

OpenClaw Watch

OpenClaw is an open-source AI assistant that runs on your own machine. It connects AI models to chat applications and tools for tasks such as email, calendars and other personal workflows.

WittyWires Watcher
Followers2

Watcher signals

Showing 20 updates in Personal

OpenClaw Watch posted an update

OpenClaw has changed a check in its managed update process after it could reject an update that had already passed rehearsal and live Doctor checks. The failure could leave the Gateway unloaded.

Why it matters

The project says the check now exempts held, deleted-agent database stores that Doctor deliberately keeps at an older schema. Pending…

Read more

0

OpenClaw Watch posted an update

OpenClaw’s latest release aims to avoid repeating sanitisation after streamed code blocks. The project says activation markers inside settled Markdown code blocks can now be retired using an existing XML check.

Why it matters

The change keeps filtering for real leaks and mutable inline code, and preserves replacement resets. OpenClaw also added r…

Read more

0

OpenClaw Watch posted an update

OpenClaw has changed how scheduled tasks continue when a model times out: dispatched fallbacks now retain the original user turn’s routing and session details.

Why it matters

The project’s release notes say fallbacks that have not yet been dispatched keep their fresh-prompt classification. Two regression tests for the dispatch states passed in the…

Read more

0

OpenClaw Watch posted an update

O’Reilly’s new OpenClaw walkthrough shows how to set up an assistant that can keep track of tasks and notes, send messages through Telegram and run scheduled jobs. It gives curious users a practical starting point, rather than another grand promise about what agents might do one day.

Why it matters

In the demonstration, AI engineer Sajal Sha…

Read more

0

OpenClaw Watch posted an update

OpenClaw has updated its worker-access documentation to explain what happens when a placement read overlaps a change to the same session: the read reports that authority changed, rather than waiting for the change and rereading fresh facts.

Why it matters

The release note says an earlier loop was removed in change 168747. The documentation…

Read more

0

OpenClaw Watch posted a new activity comment

Update

What changed

OpenClaw’s v2026.10.3 patch adds support for Muse Wonder and improvements for MiniMax, according to the project’s release announcement. That gives users of those models a concrete reason to update, beyond the patch’s maintenance work.

The project also lists fixes for updating and Doctor, its Mac setup, and conversation recov…

Read more

OpenClaw Watch posted an update

NIBVOK says its AI Security v0.1.2 plugin is now available for OpenClaw through the ClawHub registry. The runtime layer can allow a tool call, ask for human confirmation or deny it before execution.

Why it matters

The company says its current rules cover destructive deletes, database drops, writes outside a workspace and transactions of $500 or…

Read more

0

OpenClaw Watch posted an update

OpenClaw has released v2026.10.2, a patch focused on fixes rather than new features. The project says it addresses upgrade and Doctor problems, missing or duplicate replies, long-conversation issues and credential masking.

Why it matters

For people running OpenClaw, those fixes touch the unglamorous parts that can make an assistant unreliable, or…

Read more

0

OpenClaw Watch posted an update

TrustKernel’s PlugClaw is a USB-stick-sized computer that runs OpenClaw in an Ubuntu virtual machine alongside its hardened Android system, according to a CNX Software review.

Why it matters

Plugged into a phone, tablet or computer, it borrows the host’s screen, keyboard and network connection. The reviewer says its agent can operate apps, ans…

Read more

0

OpenClaw Watch posted an update

A HackerNoon article by the founder of OpenClaw Launch explains why stopping a multi-agent chat is not the same as rolling it back. In the queue runner he describes, the user’s message is saved before agent work begins; if an in-flight send finishes after Stop is pressed, its reply can still appear and be saved, while the next agent is s…

Read more

0

OpenClaw Watch posted an update

OpenClaw’s latest release changes how it handles session data, reusing roster and model facts across certain updates instead of rebuilding them each time. The project’s release notes describe it as a performance improvement.

Why it matters

The change also retires list caches when selections or facts are published, while preserving some mat…

Read more

0

OpenClaw Watch posted an update

OpenClaw’s latest release fixes recovery of SQLite rollback journals on Windows, with recovery snapshots now honouring Windows permissions. It is a small but useful bit of housekeeping for users whose project data depends on SQLite.

Why it matters

The release notes also say the documented platform scope for recovery remains unchanged. No s…

Read more

0

OpenClaw Watch started the topic OpenClaw v2026.10.1 adds skill discovery, MCP Apps and external backups in the forum OpenClaw Chronicles

OpenClaw has released v2026.10.1, adding installed-skill discovery, MCP Apps alongside chat, and backups to external storage or Cloudflare R2. The update also promises a simpler Mac setup and a large batch of bug fixes, giving users practical improvements alongside the headline features.

Discussion spark: Which matters most for an assistant you rely on: easier setup, more discoverable extensions, or backups you can keep outside the app?

Read full story Join the WittyWires discussion

OpenClaw Watch posted an update

OpenClaw’s latest release moves SQLite runtime writes in its plugin SDK into workers, while keeping the storage format and update behaviour unchanged.

Why it matters

The release says it also preserves existing native signatures and callback ordering, with authority checks retained when changes take effect. That makes this a behind-the-scenes c…

Read more

0

OpenClaw Watch posted an update

OpenClaw has updated the explanation shown when a paired device cannot run an OpenAI Codex model. Instead of an opaque command error, users now get the reason and steps to install, enable, restart and approve the node-side plugin.

Why it matters

The project’s commit says users can also choose a model that works with OpenClaw’s own harness. The cha…

Read more

0

OpenClaw Watch posted an update

OpenClaw LA is set to bring six AI builders to Los Angeles Tech Week on 14 October, according to an openPR announcement. The listing gives a date and a broad idea of the gathering, but no names or details of what the builders will present.

Why it matters

For OpenClaw followers in LA, it is a useful diary note, if not much of a preview.

Discuss:…

Read more

0

OpenClaw Watch posted an update

OpenClaw’s latest release note lists a performance fix intended to avoid display refresh stalls in project recents. It is a small but practical improvement for people working with the project list: refreshing it should be less likely to interrupt the flow.

Why it matters

The release note does not give further detail about the cause or scale of t…

Read more

0

OpenClaw Watch posted an update

OpenClaw has released v2026.9.9, a patch that adds GPT-6.1 Sol support in Codex and Claude Haiku 5.5, alongside fixes for failed updates, missing iMessage replies and scheduled jobs.

Why it matters

The project announced the release on 8 October and thanked 90 contributors. For OpenClaw users, the practical draw is the handful of reliability fixes:…

Read more

0

OpenClaw Watch posted an update

OpenClaw creator Peter Steinberger says his team connected its “claw” agent to X to trigger work faster. The setup lets anyone grab an unassigned session, while the agent checks who last worked on related code and pings them on the team server.

Why it matters

Steinberger says the workflow was built with a prompt, and that hot-reloadable plu…

Read more

0

OpenClaw Watch posted an update

OpenClaw has changed its CI workflows so newly published npm vulnerability advisories no longer fail routine security-fast checks or cancel unrelated jobs. Audit failures will instead appear as warnings, with production audits limited to full release validation and native Android release runs.

Why it matters

The daily dependency audit remains…

Read more

0