Repository navigation
chore(agent): adopt the orchestration V2 provider runtime - #1678
Conversation
The vite-hub/t3code fork rebuilt @t3tools/provider-runtime on the upstream orchestration V2 adapters after upstream deleted the V1 adapters. The host contract is unchanged, so this bumps the catalog pin and lockfile. Resume cursors have a new format; stored cursors from the previous runtime start a fresh provider thread instead of resuming.
|
Pullfrog stalled The agent stopped emitting events for 120s and was killed by the activity-timeout watchdog. The model produced no output at all before the stall — the request was sent but nothing came back. This is usually transient; re-running often succeeds. Recent agent stderr
|
Agent stopped: Pull request is no longer open. Previous results
|
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
@vite-hub/agent
@vite-hub/auth
@vite-hub/blob
@vite-hub/box
@vite-hub/browser
@vite-hub/channels
@vite-hub/cli
@vite-hub/connections
@vite-hub/content
@vite-hub/database
@vite-hub/email
@vite-hub/env
@vite-hub/kv
@vite-hub/markdown-template
@vite-hub/queue
@vite-hub/rate-limit
@vite-hub/realtime
@vite-hub/runtime
@vite-hub/sandbox
@vite-hub/schedule
@vite-hub/shell
@vite-hub/source
@vite-hub/ui
vite-hub
@vite-hub/workflow
@vite-hub/workspace
commit: |
|
Warning Your OpenAI-compatible provider endpoint is timing out, causing every run to stall before the agent can work. The last 5 Pullfrog runs on vite-hub have all failed, so this review did not happen. The model endpoint at ai.onmax.me accepts requests but returns no response before the 120-second timeout, which aborts each run. This happens consistently across multiple attempts, suggesting the provider is down, rate-limiting, or silently rejecting your CLIPROXY_API_KEY. To fix it:
|
Provider Agents embed
@t3tools/provider-runtimefrom the vite-hub/t3code fork. Upstream T3 Code merged orchestration V2 and deleted the V1 Claude and Codex adapters the runtime wrapped, so the fork stopped syncing on 2026-09-25 and the pinned package fell a month behind upstream provider fixes.The fork rebuilt the runtime on the V2 adapters and kept the host contract (vite-hub/t3code#6). This change bumps the catalog pin and lockfile to that fork commit. No vitehub code changes were needed:
packages/agenttypechecks against the new types and the mocked Provider Agent tests pass unchanged.Behaviour to know about:
steerinput.ProviderRuntimeEventshape. Provider retries arrive as recoverableruntime.warningevents; the run outcome comes from the provider turn terminal.Validation in the task worktree with the published preview packages:
pnpm installresolved@t3tools/provider-runtimeand@t3tools/contractsfrom the new fork commit.vp run -t @vite-hub/agent#buildtsc --noEmit -p packages/agent/tsconfig.jsonvp testforprovider-agent,provider-runtime-packages,providers,provider-status,provider-toolchain,provider-driver-worker,nuxt-runtime-output,box,eval, andbabysitter-preset(all pass; three hard-link tests need a temp directory on the same filesystem as the checkout).Not verified: a live Codex or Claude turn through the new runtime.
Model: Claude Fable 5.1
Harness: T3 Code (Claude Code)