OpenClaw apologizes for messy release and promises an LTS

OpenClaw is an open-source infrastructure project built to bridge Discord, Telegram, WhatsApp, and other chat channels with backend AI models and developer tools. On May 5, the team issued a formal apology after version 2026.4.29 triggered widespread stability issues, outlining plans to slim down its core and launch a dedicated LTS release.

The trouble first surfaced around version 2026.4.24. By 2026.4.29, gateways slowed down significantly, certain self-hosted instances fell into infinite plugin dependency repair loops, and messaging performance across Discord, Telegram, and WhatsApp degraded. Many users had to downgrade their setups to restore normal operations.

How a refactor went wrong

The downtime was not caused by a single isolated bug, but by a messy mid-flight architectural shift.

Recent supply-chain security incidents across the npm ecosystem, including the Axios compromise, highlighted the severe risks lurking in deep dependency trees and install-time scripts. OpenClaw did not directly depend on Axios, but to reduce dependency graph risks, the team began stripping optional integrations, heavy parsers, providers, and chat channels out of core and moving them to ClawHub.

The transition landed in the worst possible middle state. Plugin dependency repairs ran directly inside startup and update paths, internal and external plugins were half-separated, ClawHub artifact metadata was still settling, and gateway cold paths were saddled with too much work.

The episode also exposed an operational bottleneck. OpenClaw had remained too founder-driven, with releases, reviews, packaging, and support resting on one person. Through the OpenClaw Foundation, the project is now expanding into a proper engineering team with help from OpenAI.

What is changing next

The team is clarifying plugin boundaries to show exactly what remains bundled in core, what installs separately, and what requires a source checkout.

Release engineering is also getting restructured. Alongside its regular fast-paced update cycle, OpenClaw plans to announce an LTS (Long Term Support) release in late May for users running critical infrastructure.

If you are currently running OpenClaw in production, the safest move is to stick with an earlier stable release or downgrade, holding off on upgrades until the late May LTS arrives.

Why it matters

For developers building self-hosted, multi-channel AI agents, OpenClaw is a versatile tool, but this bumpy week shows it is still maturing out of its experimental phase. If you rely on it for reliable daily workflows, you should avoid tracking bleeding-edge updates and wait for the late May LTS release instead.

Compared to lightweight wrappers that ignore supply-chain bloat, OpenClaw’s push to shrink its core and untangle deep dependencies is the right technical call. Evolving from a founder-run project into an OpenAI-backed foundation is a painful transition, but it is necessary if the tool wants to become durable infrastructure.