Repository navigation
First public release of the community TanStack AI adapter for OpenCode v2 hosted inference. It calls the inference APIs directly from your server and requires no OpenCode CLI process.
npm install @tanstack/ai tanstack-ai-opencode-consoleRequires Node.js 22.19+ and @tanstack/ai 0.65.x.
Included in this release:
- Chat Completions, Responses, Anthropic Messages, and Gemini through native TanStack protocol adapters.
- Service-account keys and workspace-scoped Console device sign-in, with session refresh helpers and server-side credentials.
- Go subscription inference via an explicit gateway root and protocol overrides.
- A private React/Vite chat example with streaming text, reasoning, a server time tool, and cancellation.
Validation: 104 fixture tests, builds, types, formatting, and isolated package-consumer checks passed on Node 22, 24, and 26. Separate live Console and Go checks are documented in the authentication evidence and Go guide.
The attached .tgz is the exact npm 0.1.0 archive, verified against the registry SHA-512 integrity and SHA-1 checksum. SHA256SUMS supports verification of the GitHub download:
sha256sum -c SHA256SUMSSource tag: tanstack-ai-opencode-console@0.1.0, commit 76f82b6b6c963a27ab1c947864aecb30019934cb.
This bootstrap release was published locally without provenance. The package retains its provenance configuration for future CI releases.