feat(models): add Azure AI Foundry to unified model provider setup#13924
Closed
pareek-ml wants to merge 2 commits into
Closed
feat(models): add Azure AI Foundry to unified model provider setup#13924pareek-ml wants to merge 2 commits into
pareek-ml wants to merge 2 commits into
Conversation
Adds Azure AI Foundry as a first-class model provider in Langflow's unified model system, enabling centralized configuration via Settings → Model Providers. - Register "Azure AI Foundry" in MODEL_PROVIDER_METADATA with AZURE_AI_FOUNDRY_API_KEY (required, secret, langchain_param=credential) and AZURE_AI_FOUNDRY_ENDPOINT (required, non-secret, langchain_param=endpoint) - Add AzureAIOpenAIApiChatModel to the lazy-import class registry (langchain-azure-ai package, with install hint) - Add seed model catalog (gpt-4o, gpt-4o-mini, gpt-4.1, o3-mini, Mistral-Large-3) with capability flags to _STATIC_MODELS_DETAILED - Extend validate_model_provider_key with an Azure AI Foundry branch that constructs AzureAIOpenAIApiChatModel and invokes a test call; returns early when api_key, endpoint, or model_name is absent - Extend get_llm with an Azure AI Foundry branch that resolves AZURE_AI_FOUNDRY_ENDPOINT from provider vars or env and raises ValueError with an actionable message when missing - Register both env vars in VARIABLES_TO_GET_FROM_ENVIRONMENT for automatic environment import - Add langchain-azure-ai to flow requirements map so flows using this provider include the package in generated requirements.txt Tests (12 total, full coverage of new paths): - Provider registry presence and metadata shape - Parameter mapping resolution (model_class, api_key_param) - Env var registration and requirements generation - Credential validation: success path and all three early-return paths (missing model, missing api_key, missing endpoint) - get_llm: endpoint + credential wiring and missing-endpoint error Supersedes langflow-ai#13912 with docstrings on all test functions (fixing CodeRabbit's docstring-coverage warning) and tests for the three credential validation early-return paths (fixing Codecov's credentials.py partial-coverage report).
Contributor
|
Important Review skippedAuto incremental reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
WalkthroughThis PR adds Azure AI Foundry as a unified model provider in lfx. It introduces model catalog metadata, provider configuration, a lazy-import class registry entry, credential validation logic, LLM instantiation wiring (endpoint resolution with gated env fallback), requirement-package fallback mapping, environment variable registration, and accompanying unit tests. ChangesAzure AI Foundry provider integration
Sequence Diagram(s)sequenceDiagram
participant Flow
participant get_llm
participant _env_if_allowed
participant AzureAIOpenAIApiChatModel
Flow->>get_llm: request LLM("Azure AI Foundry", model_name)
get_llm->>get_llm: resolve AZURE_AI_FOUNDRY_ENDPOINT (component/db)
get_llm->>_env_if_allowed: check env fallback if not found
_env_if_allowed-->>get_llm: endpoint value or None
alt endpoint missing
get_llm-->>Flow: raise ValueError("Azure AI Foundry endpoint is required")
else endpoint resolved
get_llm->>AzureAIOpenAIApiChatModel: construct(credential, endpoint, model)
AzureAIOpenAIApiChatModel-->>get_llm: chat model instance
get_llm-->>Flow: return LLM instance
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~30 minutes Possibly related PRs
Suggested labels
Suggested reviewers
🚥 Pre-merge checks | ✅ 8 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (8 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
github-actions
Bot
added
enhancement
Author
|
Closing in favour of the cleaner bundle approach. After looking at how #13916 works, this PR is doing exactly what #13916 was built to prevent — touching Most of those core edits are already handled generically by the registry:
The one genuine gap is The right fix is a small extension to models: list[dict] = field(default_factory=list)…and a corresponding append to I'll open a PR against cc @erichare |
github-actions
Bot
added
enhancement
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
🧹 Nitpick comments (1)
src/lfx/tests/unit/base/models/test_azure_ai_foundry_provider.py (1)
150-169: 🩺 Stability & Availability | 🔵 Trivial | ⚡ Quick winEarly-return tests don't guard against a real network call on regression.
Unlike the success test, these tests don't mock
langchain_azure_ai/AzureAIOpenAIApiChatModel, yetmodel_name="gpt-4o"is supplied. If the early-return guard incredentials.pyever regresses, these tests would attempt a real import/network call to Azure instead of failing with a clear assertion — making CI flaky/slow instead of deterministically red.🧪 Suggested fix: mock the model and assert it wasn't called
def test_validate_model_provider_key_azure_ai_foundry_returns_early_without_api_key(): """validate_model_provider_key must return without error when API key is absent.""" from lfx.base.models.unified_models import validate_model_provider_key - validate_model_provider_key( - "Azure AI Foundry", - {"AZURE_AI_FOUNDRY_ENDPOINT": "https://example.services.ai.azure.com/openai/v1"}, - model_name="gpt-4o", - ) + calls = [] + + class FakeFoundryChatModel: + def __init__(self, **kwargs): + calls.append(kwargs) + + fake_chat_models = SimpleNamespace(AzureAIOpenAIApiChatModel=FakeFoundryChatModel) + with patch.dict( + "sys.modules", + {"langchain_azure_ai.chat_models": fake_chat_models}, + ): + validate_model_provider_key( + "Azure AI Foundry", + {"AZURE_AI_FOUNDRY_ENDPOINT": "https://example.services.ai.azure.com/openai/v1"}, + model_name="gpt-4o", + ) + assert calls == []Same applies to
test_validate_model_provider_key_azure_ai_foundry_returns_early_without_endpoint(Lines 161-169).As per coding guidelines, "verify the tests actually cover the new or changed behavior rather than acting as placeholders."
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/lfx/tests/unit/base/models/test_azure_ai_foundry_provider.py` around lines 150 - 169, The Azure AI Foundry early-return tests in test_validate_model_provider_key_azure_ai_foundry_returns_early_without_api_key and test_validate_model_provider_key_azure_ai_foundry_returns_early_without_endpoint should also mock the underlying langchain_azure_ai / AzureAIOpenAIApiChatModel path so they fail deterministically if the guard in validate_model_provider_key (from unified_models/credentials logic) regresses. Update both tests to patch the model import/construction and assert it is not called when either the API key or endpoint is missing, instead of relying only on the absence of an exception.Source: Coding guidelines
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Nitpick comments:
In `@src/lfx/tests/unit/base/models/test_azure_ai_foundry_provider.py`:
- Around line 150-169: The Azure AI Foundry early-return tests in
test_validate_model_provider_key_azure_ai_foundry_returns_early_without_api_key
and
test_validate_model_provider_key_azure_ai_foundry_returns_early_without_endpoint
should also mock the underlying langchain_azure_ai / AzureAIOpenAIApiChatModel
path so they fail deterministically if the guard in validate_model_provider_key
(from unified_models/credentials logic) regresses. Update both tests to patch
the model import/construction and assert it is not called when either the API
key or endpoint is missing, instead of relying only on the absence of an
exception.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 3ef7e3ed-bcca-474e-bfbc-bfd3cb20aba4
📒 Files selected for processing (9)
src/lfx/src/lfx/base/models/azure_ai_foundry_constants.pysrc/lfx/src/lfx/base/models/model_metadata.pysrc/lfx/src/lfx/base/models/unified_models/class_registry.pysrc/lfx/src/lfx/base/models/unified_models/credentials.pysrc/lfx/src/lfx/base/models/unified_models/instantiation.pysrc/lfx/src/lfx/base/models/unified_models/provider_queries.pysrc/lfx/src/lfx/services/settings/constants.pysrc/lfx/src/lfx/utils/flow_requirements.pysrc/lfx/tests/unit/base/models/test_azure_ai_foundry_provider.py
pareek-ml
added a commit
to pareek-ml/langflow
that referenced
this pull request
Jul 3, 2026
…ai-foundry bundle Part 1 — registry extension (provider_registry.py, manifest.py, loader): Adds a `models: list[dict]` field to ProviderSpec so a bundle can seed a static model catalog without editing core files. register_provider() appends the list to _STATIC_MODELS_DETAILED in-place; clear() reverses it. The extension.json schema gains a matching `models[]` array; the loader passes it through to ProviderSpec. Zero behavior change when no bundle is loaded. Part 2 — lfx-azure-ai-foundry bundle (src/bundles/azure-ai-foundry/): Ships Azure AI Foundry as a provider-only Extension Bundle with zero core edits, superseding the core-edit approach in langflow-ai#13912 and langflow-ai#13924. - extension.json: declares provider metadata (AZURE_AI_FOUNDRY_API_KEY + AZURE_AI_FOUNDRY_ENDPOINT), AzureAIOpenAIApiChatModel lazy import, dotted- path validator, and a seed catalog (gpt-4o, gpt-4o-mini, gpt-4.1, o3-mini, Mistral-Large-3) via the new models[] field. - validator.py: constructs AzureAIOpenAIApiChatModel and calls invoke(); early- returns when api_key, endpoint, or model_name is absent. Tests (10): - models[] injection into _STATIC_MODELS_DETAILED and clear() reversal. - End-to-end load through the real extension loader: provider appears in get_model_providers() and seed models in get_models_detailed(). - Validator: early-return paths (missing model, key, endpoint), success, and ValueError on invoke() failure. Refs langflow-ai#13912, closes langflow-ai#13924
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.
What & why
Adds Azure AI Foundry as a first-class model provider in Langflow's unified model system (Settings → Model Providers), enabling centralized configuration of endpoint and API key — identical to how OpenAI, Anthropic, and Ollama work today.
Supersedes #13912 by @HimavarshaVS with two fixes applied:
missing model_name,missing api_key,missing endpoint) are now tested (fixes Codecov'scredentials.pypartial-coverage report)How
azure_ai_foundry_constants.pygpt-4o,gpt-4o-mini,gpt-4.1,o3-mini,Mistral-Large-3with capability flagsmodel_metadata.py"Azure AI Foundry"entry:AZURE_AI_FOUNDRY_API_KEY(credential) +AZURE_AI_FOUNDRY_ENDPOINT(endpoint),AzureAIOpenAIApiChatModelmappingclass_registry.pyAzureAIOpenAIApiChatModelfromlangchain-azure-aiprovider_queries.py_STATIC_MODELS_DETAILEDcredentials.pymax_tokens=1, invokes test call; early-returns when any required field is absentinstantiation.pyget_llmbranch: resolvesAZURE_AI_FOUNDRY_ENDPOINTfrom provider vars or env; raisesValueErrorwith actionable message when missingsettings/constants.pyVARIABLES_TO_GET_FROM_ENVIRONMENTflow_requirements.py"Azure AI Foundry": {"langchain-azure-ai"}Tests
12 tests in
test_azure_ai_foundry_provider.pycovering:get_llm: endpoint + credential wiring and missing-endpointValueErrorTest plan
gpt-4o)Refs #13912
Summary by CodeRabbit
New Features
Bug Fixes
Tests