Alibaba Cloud's open-source autonomous agent framework, HiClaw, has unveiled its v1.1.0 release, bringing significant enhancements to its architecture and functionality. This update introduces native Kubernetes cluster deployment capabilities and comprehensive support for the Hermes Worker runtime, marking a pivotal advancement in how autonomous agents can be deployed and managed within enterprise environments.
The v1.1.0 release fundamentally refactors HiClaw's architecture by transitioning from a single — container mode to a standard Controller — Reconciler architecture built directly on the Kubernetes control plane. This strategic shift leverages Kubernetes' robust orchestration capabilities, allowing HiClaw to run on top of the native control plane. The `hiclaw — controller` now coordinates Worker, Team, Manager, and Human Custom Resources (CRs) into containers, Matrix rooms, and gateway routes. For development and smaller deployments, an embedded mode featuring a lightweight `kube-apiserver` and `kine` for CRD data eliminates the need for an external Kubernetes cluster.
A cornerstone of this update is the integration of the Hermes Worker runtime, now a first — class citizen designed specifically for autonomous programming tasks. Unlike the existing `agent` (Node.js) and `QwenPaw` (Python) runtimes, which primarily handle conversations and tool calls, Hermes is an autonomous programming Agent capable of independently planning, executing, and iterating on complex software development tasks. Its robust capabilities include terminal sandbox execution, multi — file code generation, debugging, visual analysis, and native mautrix Matrix integration, all operating within isolated containers.
To further enhance enterprise readiness, HiClaw v1.1.0 provides a production — grade Helm Chart for deploying the framework on Kubernetes clusters. This Chart deploys essential services such as Tuwunel (Matrix server), MinIO (object storage), Element Web (IM client), and the `hiclaw — controller` as separate Deployments or StatefulSets, complete with the necessary Service, RBAC, and Secret resources. Key enterprise features enabled by this deployment model include Leader Election, which supports multi — replica controller deployments with automatic failover for high availability.
The update also introduces a more flexible and efficient multi — container architecture and pluggable provider interfaces. The Controller now delegates gateway operations (via Higress) and storage operations (MinIO/OSS) through modular Provider interfaces, enabling easier integration with various backends. A new `hiclaw — credential-provider` Sidecar specifically handles STS token issuance, key rotation, and enforces per-worker access policies, with support for Alibaba Cloud OSS, AWS S3, or any S3-compatible storage backend without requiring changes to the Controller's core code. This architectural redesign also leads to a more streamlined `Manager` image, which no longer bundles infrastructure services like Higress, Tuwunel, MinIO, and Element Web.
Further optimizing the framework, the built — in OpenClaw engine has been upgraded to `hiclaw — 2026.4.14`, incorporating Matrix private — network security fixes, structured Matrix debug logs, and unified gateway Control UI ports. A significant achievement in this release is the substantial image slimming, which reduced the size of all downstream images — including manager, worker, copaw — worker, and hermes — worker—by approximately 1.7 GB. This was accomplished by resetting the `openclaw — base` base image from the hefty `higress/all-in-one` (~1.79 GB) to a more compact `higress/ubuntu:24.04` (~103 MB). Compatibility fixes were also implemented, such as setting `gateway.
These comprehensive enhancements collectively propel HiClaw into a new era of enterprise readiness, offering a more scalable, fault — tolerant, and versatile platform for AI agent systems. Developers and organizations can now leverage the stability and advanced features of a Kubernetes — native infrastructure, coupled with the specialized capabilities of the Hermes runtime, to deploy and manage complex autonomous programming agents with greater efficiency and robustness.
Sources
Replies (0)
No replies in this topic yet.
