v0.23.0-beta.4

Pre-release June 4, 2026

Netclaw 0.23.0-beta.4

Netclaw v0.23.0-beta.4 — streaming-native chat stack, composable routing, and spinner standardization

This is a prerelease on the beta channel. Stable installs are unaffected and remain on 0.22.1; only opt-in beta testers are offered this build.

Try the beta

Linux / macOS: curl -sSL https://releases.netclaw.dev/install.sh | bash -s -- --channel beta
Windows: download install.ps1, then ./install.ps1 -Channel beta
Docker: docker pull ghcr.io/netclaw-dev/netclaw:beta

Features

Streaming-native chat-client stack with composable routing — the IChatClient stack has been redesigned so the streaming-vs-non-streaming transport distinction no longer leaks into callers. Netclaw now issues only streaming requests across all 8 providers, eliminating the OpenAI Codex 400 "Stream must be set to true" error and preventing reasoning content drops on some providers. Resilience and observability are compositional via Microsoft.Extensions.AI.ChatClientBuilder — each (provider, model) pipeline is assembled as Logging → Retry → VendorOptions → raw. A RoutingChatClient walks ordered candidate lists for failover, and LoggingChatClient is now stateless with session-correlated log tags (SessionId). The transport-level retry policy is configurable via Session:Tuning:StreamingRetryPolicy, replacing the previous overlapping actor-level retry. (#1313)

Bug Fixes

Standardized self-animating spinner across probe surfaces — replaced five hand-rolled, drifted probe/validation spinners with Termina's shared SpinnerNode helper. Fixes the frozen ModelManager spinner, the ~1fps crawl on provider validation/probe spinners, and inconsistent animation cadence. Views now just drop in the spinner node — no per-surface tick field or hand-wired redraw subscription. (#1312, #1327)