Add xAI as a commit message provider - #34
Merged
Merged
Conversation
- Rename `schema_OpenAITokens` -> `schema_BearerTokens` and `OpenAITokens` -> `BearerTokens`, updating importers in `src/infra/auth/openai.ts` and `src/infra/storage/config.ts`, since the shape is not OpenAI-specific. - Replace `updateGoogleTokens` and `updateOpenAITokens` with a single `updateOAuthTokens` taking the tagged `RefreshableAuthMethod` variant, so the tag and payload cannot disagree and the two exhaustive `auth_method.type` switches disappear. - Derive `RefreshableAuthMethod` structurally from `RefreshTokens` so it widens on its own when a refreshable auth variant is added. - Pass the tagged variant from `resolveProvider` in `src/domain/llm/auth-resolver.ts`. - Replace the `google_oauth || openai_oauth` boolean chain in `src/cli/doctor.ts` with an exhaustive `tokenExpiry` helper returning `Maybe<number>`, so a new auth variant is a compile error rather than a silently missing Token Status row. - Return `absurd` from the `authMethodLabel` default instead of a placeholder string, for the same reason. - Delete the unused `AI_PROVIDERS` export, which had no importers and did not drive the setup picker. - Cover `updateOAuthTokens` with tests for the matching, mismatched, and sibling-field-preservation cases, which neither original updater had.
- Add the `xai` variant to `schema_ProviderConfig` with `XAI_EFFORTS` (`low`, `high`), checked with `satisfies` against the OpenAI SDK's own reasoning effort type. - Add `src/infra/auth/xai.ts` exposing `xaiApiKeyOptions`, which points the `openai` client at `https://api.x.ai/v1`; xAI's API is OpenAI-compatible so no new SDK is needed. - Add `src/infra/llm/xai.ts` calling `chat.completions` non-streaming, since the response carries both the message and usage in one payload. - Retry once without `reasoning_effort` when Grok rejects it, because support is per-model and `/v1/models` does not advertise it; skip the retry when no effort was sent. - Route `xai` through `generateContent` and report its effort as `provider default`, matching what the client reports after an effort-stripping retry. - Add `xai` arms to `seedProviderConfig`, `withModel`, `selectEffortForProvider`, `selectXaiEffort`, and `fetchModels`. - Offer xAI in `commit setup` and label its key in `commit doctor`. - Script the `@clack/prompts` `select` mock per test in `test/cli/setup.test.ts` so wizard tests no longer share one positional chain. - Cover the xAI client, model listing, router dispatch, config round-trip, and setup wizard with tests. - Document xAI in the setup prompts and providers list in `README.md`.
- Add the `xai_oauth` auth variant carrying `schema_BearerTokens`, so a SuperGrok or X Premium subscription can drive the provider instead of a metered API key. - Add the authorization-code + PKCE S256 flow in `src/infra/auth/xai.ts` against `https://auth.x.ai`, with endpoints read once from the OIDC discovery document and hardcoded, since refresh runs on every command and discovery would add a round-trip to each one. - Bind the loopback callback server on an OS-assigned port and read the bound port back, rather than reserving a fixed one; the redirect URI is therefore built inside the bracket, once the server is listening. - Point OAuth requests at `https://cli-chat-proxy.grok.com/v1` with `X-XAI-Token-Auth: xai-grok-cli`, which is what marks the bearer as a user token rather than a deployment key. - Set `maxRetries: 0` on the proxy client so a 429 against a metered subscription is not silently retried three times; `withTransientRetry` already owns retry policy. - Extract `generateCodeVerifier`, `generateCodeChallenge`, `generateState`, `stopCallbackServer`, `openBrowser`, and `oauthTimeout` into `src/infra/auth/oauth.ts`, shared by the OpenAI and xAI flows. - Refresh and persist xAI tokens through `resolveProvider`, and report their expiry in `commit doctor`. - Offer "Sign in with Grok" as the default xAI auth method in `commit setup`. - Add `xai_oauth` arms to the auth switches in the Gemini, OpenAI, and Anthropic clients and in `fetchModels`. - Cover the authorize URL, PKCE derivation, refresh buffer, token rotation, and revoked-token message with tests, plus the proxy client options for chat and model listing. - Document the subscription sign-in in `README.md`.
- Add `x-grok-client-version` to the OAuth client headers; `/chat/completions` on the proxy returns HTTP 426 without it and enforces a server-side minimum version. - Take the value from `xai-grok-pager` in xai-org/grok-build, and record in `XAI_CLIENT_VERSION` that it must be bumped whenever xAI raises the floor. - Map a 426 to a message naming `XAI_CLIENT_VERSION`, since xAI's own text tells the user to run `grok update`, which does not apply to this CLI. - Re-export `APIError` from the `openai` mock, which the new status check needs. - Cover the version header and the 426 message with tests.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 8b05261493
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
- Keep only `grok-` ids that do not include `-image` in `fetchXaiModelsWith`, matching the OpenAI chat-only prefix filter so setup cannot persist image models. - Cover image and non-grok exclusions with a unit test on the xAI catalog path.
Owner
Author
|
@codex review — focus only on P0/P1: security, correctness, broken contracts, auth, data loss. Skip P2/P3 style, nits, and speculative edge cases. Report only major issues or say you didn't find any major issue. |
|
Codex Review: Didn't find any major issues. Another round soon, please! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Motivation
Just adding a new provider.
What's New
OAuth Token Persistence
updateOAuthTokensupdater for any OAuth auth method.xAI Provider (API Key)
xaiprovider, models, and effort options in config and the commit model catalog.generateContentWithXaitohttps://api.x.ai/v1via the OpenAI SDK.reasoning_effortwhen the model rejects it.Grok Subscription OAuth
auth.x.ai, token refresh, andxai_oauthconfig variant.cli-chat-proxy.grok.comwithX-XAI-Token-AuthandmaxRetries: 0.x-grok-client-versionso the proxy accepts completions (maps 426 to a local bump hint).CLI and Docs
Tests
xAI Auth and Completion Flow
graph TD Setup[commit setup] --> Pick{Auth method} Pick -->|API key| ApiKey[Store api_key] Pick -->|Grok OAuth| OAuth[PKCE via auth.x.ai] OAuth --> Tokens[Store xai_oauth tokens] ApiKey --> Commit[generateCommitMessage] Tokens --> Refresh[ensureFreshXaiTokens] Refresh --> Commit Commit --> Xai[generateContentWithXai] Xai --> Auth{auth_method} Auth -->|api_key| Public[api.x.ai/v1] Auth -->|xai_oauth| Proxy[cli-chat-proxy.grok.com + client version] style Xai fill:#808080,stroke:#333,stroke-width:2px style Proxy fill:#808080,stroke:#333,stroke-width:2pxTesting & Feedback
commit setup, choose xAI with API key, and generate a commit message against a real diff.reasoning_effort.If you find any bugs or have recommendations for improvements, please open an issue and assign it to me.