[Taranum01]

"Fixes #3682.\n\n## Problem\n\nConfiguring a credential for an OpenAI-API-compatible custom model in a Knowledge Pipeline failed with \"does not have provider_credential_schema\" \u2014 the form could not render and the user saw a red error toast.\n\nDify requires every provider used in a Knowledge Pipeline to declare a top-level provider_credential_schema so the authorization form can render. The OpenAI-API-compatible plugin only declared model_credential_schema, so the provider-level schema was missing. (Other configurable-model plugins like openai, anthropic, gemini, mistral, etc. all have it.)\n\n## Fix\n\nAdded the missing provider_credential_schema with the two universal fields every OpenAI-compatible server needs:\n\n| Field | Type | Required | Notes |\n| --- | --- | --- | --- |\n| api_key | secret-input | yes | The bearer token for the provider. |\n| endpoint_url | text-input | no | The OpenAI-compatible server URL. Empty falls back to the default OpenAI endpoint; set to point at LM Studio / vLLM / LiteLLM / etc. |\n\nThe model-level credential_form_schemas is preserved unchanged, so existing per-model configuration flows (LLM / Embedding / Rerank / Speech2Text / TTS) are not affected. The configurate_methods list still contains only customizable-model, so the global provider credential form is the only new entry point.\n\n## Tests\n\nmodels/openai_api_compatible/tests/test_provider_credential_schema.py (5 tests):\n\n- test_provider_yaml_declares_provider_credential_schema \u2014 the whole point of the fix.\n- test_provider_credential_form_has_api_key \u2014 pins type (secret-input), required flag, and bilingual label.\n- test_provider_credential_form_has_optional_endpoint_url \u2014 pins type (text-input), required flag = False.\n- test_model_credential_schema_preserved \u2014 guards against accidentally dropping the per-model schema during the refactor.\n- test_provider_yaml_round_trip \u2014 PyYAML safe_load round-trip, catches accidentally using !!python/str or other unsafe tags.\n\n\n$ uv run pytest tests/test_provider_credential_schema.py -v\n========================= 5 passed, 2 warnings in 0.11s ==========================\n\n\n$ uv run pytest tests/ \u2014 93 passed (88 pre-existing + 5 new). ruff check and ruff format --check clean on the new file.\n\n## Release Notes\n\n- fix(openai_api_compatible): add the top-level provider_credential_schema so the Knowledge Pipeline credential form can render. Previously, configuring a credential for an OpenAI-API-compatible custom model in a Knowledge Pipeline failed with "does not have provider_credential_schema" \u2014 the form could not render and the user saw a red error toast. The new schema adds api_key (required) and endpoint_url (optional) form fields. The per-model credential_form_schemas is preserved unchanged. Bumps the plugin from 0.0.66 \u2192 0.0.68.\n"

[crazywoola]

Thank you for your contribution, @Taranum01! This PR currently has merge conflicts with main, so we are closing it for now. Please update your branch against the latest main and resolve the conflicts — note that the consolidated dependency refresh (#4106–#4111) has bumped plugin manifest.yaml versions on main, so please bump models/openai_api_compatible/manifest.yaml to the next unused version when you rebase. You are welcome to open a new PR with the updated changes and link back to this one. We appreciate your time and effort!

If these changes are still wanted, they can be folded into #3795 when you update it.