feat: move mfb provider from azure to pair foundry#9426
Draft
kevin9foong wants to merge 9 commits into
Draft
Conversation
kevin9foong
force-pushed
the
feat/mfb-pair-foundry
branch
from
6668595 to
79b6906
Compare
Adds the Vercel AI SDK (`ai`) to the backend so the MFB LLM client can be migrated off the Azure OpenAI client onto `@ai-sdk/openai`'s `createOpenAI` provider pointed at Pair Foundry's PX Engine. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces the Azure OpenAI client in `ai-model.ts` with a Vercel AI SDK
wrapper built on `createOpenAI` (`@ai-sdk/openai`) plus `generateText`
(`ai`), targeting Pair Foundry's PX Engine via the existing
`aisdkConfig` (providerName, apiKey, baseUrl, modelName from env/SSM).
Behaviour-preserving from the caller's perspective:
- `sendPromptToModel({ messages, options, formId })` keeps its
signature and `ResultAsync<string | null, ...>` return type.
- Provider/client construction failures still surface as
`ModelGetClientFailureError`; request failures as
`ModelResponseFailureError` (with `formId` in log meta); empty/missing
text in the model response still returns `null`.
- `temperature` stays unpinned; callers control it via `options`.
Internal type and shape changes:
- `Message` is now ai-sdk's `ModelMessage`; the local `Role` enum is
removed (callers use the string literals `'user'`/`'system'`).
- Vision-flow image parts move from the OpenAI shape
`{ type: 'image_url', image_url: { url } }` to ai-sdk's
`{ type: 'image', image: <dataURL> }`.
- JSON-mode is plumbed per-call via
`options.providerOptions.openai.responseFormat`, replacing the
OpenAI-SDK-native `response_format` option. Downstream `JSON.parse`
+ zod validation in the assistance service is unchanged.
Existing `admin-form.assistance.service.spec.ts` continues to mock
`sendPromptToModel` wholesale and passes unchanged through the swap.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a wrapper-level spec for `sendPromptToModel` that mocks the ai-sdk boundary (`generateText`, `createOpenAI`) rather than going deeper, so it stays valid if internals are later refactored (e.g. swapping `generateText` for `generateObject` in the structured-outputs follow-up). Covers: - Happy path: messages forwarded to `generateText` with the model returned by the OpenAI provider, and `result.text` is returned. - Null response: `generateText` resolves with empty `text` → wrapper returns `ok(null)`. - Client construction failure: `createOpenAI` throws → wrapper returns `err(ModelGetClientFailureError)` and never calls `generateText`. - Request failure: `generateText` rejects → wrapper returns `err(ModelResponseFailureError)` and logs include `formId` in meta. - Options pass-through: caller-supplied `temperature` and `providerOptions.openai.responseFormat` reach `generateText`. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
kevin9foong
force-pushed
the
feat/mfb-pair-foundry
branch
from
79b6906 to
d37c0cd
Compare
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.
Problem and solution
Shift from azure AI foundry to pair foundry
This should provide the following benefits:
Closes [insert issue #]
Breaking Changes
No - this PR is backwards compatible
Tests
Deploy Notes
New environment variables:
env var: env var detailsdependency: dependency detailsai-sdk, this is the recommended approach by the pair team. for interoperabilty, the @ai-sdk/openai provider also works with any OpenAI-compatible API we wish to use in the future.