migrating-ai-sdk-to-common-ai

Migrate airflow-ai-sdk to apache-airflow-providers-common-ai

This skill migrates Airflow projects from airflow-ai-sdk to apache-airflow-providers-common-ai (target 0.4.0+), the official Airflow AI provider built on PydanticAI. It also covers upgrading projects already on common-ai 0.1.x, since several capabilities (multimodal prompts, toolsets, embedding operators, structured-output XCom behavior) changed between 0.1.0 and 0.4.0.

CRITICAL: The new provider requires Airflow 3.0+ and (for 0.4.0) pydantic-ai-slim >= 1.71.0. The API surface has changed: LLM configuration moves from code (model strings/objects) to Airflow connections (pydanticai type). There is no @task.embed in the new provider; embeddings move to the LlamaIndex integration or a plain @task (see Step 3).

Before starting

Use the Grep tool with the pattern below to inventory everything that needs to migrate:

airflow_ai_sdk|airflow-ai-sdk|ai_sdk|@task\.llm|@task\.agent|@task\.llm_branch|@task\.embed

From the results, capture:

1. All files importing airflow-ai-sdk / airflow_ai_sdk
2. Which decorators are in use: @task.llm, @task.agent, @task.llm_branch, @task.embed
3. The model configuration pattern (string names like "gpt-5", or OpenAIModel(...) objects)
4. Any airflow_ai_sdk.BaseModel subclasses used as output_type

Use this inventory to drive the steps below.

Step 1: Update requirements.txt

Remove:

airflow-ai-sdk[openai]
# or any variant: airflow-ai-sdk[openai]==0.1.7, airflow-ai-sdk[anthropic], etc.

Add:

apache-airflow-providers-common-ai[openai]>=0.4.0

Use the latest available 0.x version unless the user has pinned a specific one. Available extras (0.4.0): [openai], [anthropic], [google], [bedrock], [llamaindex], [langchain], [mcp], plus file-format extras ([pdf], [docx], [parquet], [avro]) for DocumentLoaderOperator and [sql]/[common-sql] for the SQL operators. There are no [groq]/[mistral] extras; for those providers install the matching pydantic-ai-slim extra yourself.

Add [llamaindex] if the project migrates @task.embed to the LlamaIndexEmbeddingOperator (recommended, see Step 3). In that case sentence-transformers and torch can usually be removed, which shrinks the image considerably. Keep them only if the project stays on local sentence-transformers embeddings via plain @task.

Step 2: Create PydanticAI connection

The new provider uses an Airflow connection instead of model strings or objects in code.

Connection type: pydanticai Default connection ID: pydanticai_default

Via environment variable (.env)

AIRFLOW_CONN_PYDANTICAI_DEFAULT='{
    "conn_type": "pydanticai",
    "password": "<api-key>",

Model format

The model field uses provider:model format:

Custom endpoints (Ollama, vLLM, Snowflake Cortex, etc.)

Set host to the base URL:

Use the openai: prefix for any OpenAI-compatible API, regardless of the actual provider.

Connection ID convention

The env var name determines the connection ID:

• AIRFLOW_CONN_PYDANTICAI_DEFAULT creates pydanticai_default
• AIRFLOW_CONN_PYDANTICAI_CORTEX creates pydanticai_cortex

Model resolution priority

1. model_id parameter on the decorator/operator (highest)
2. model in connection's extra JSON (fallback)

Other connection types (0.4.0)

Besides pydanticai, the provider registers vendor-specific connection types: pydanticai-azure (Azure OpenAI: host = endpoint, extra api_version), pydanticai-bedrock (AWS credentials/region in extra), and pydanticai-vertex (GCP project/location in extra). The LlamaIndex and LangChain hooks read API key/host/extra from whatever connection ID they are given, so a single pydanticai_default connection can serve LLM calls and embeddings: one API key entry for the whole project.

Step 3: Migrate decorators

@task.llm

Parameter mapping:

Multimodal prompts (0.4.0+): the translation function may return a Sequence[UserContent] instead of a string, e.g. for vision:

This matches the old airflow-ai-sdk vision pattern, so vision code migrates unchanged. Note: common-ai 0.1.x only accepted strings — if a project disabled vision to migrate to 0.1.0, re-enable it when bumping to 0.4.0. Non-string prompts are incompatible with require_approval=True / enable_hitl_review=True (both render the prompt as text).

Structured output via XCom (0.4.0 behavior change): with output_type=<BaseModel subclass>, the model instance flows through XCom on Airflow cores whose task SDK has SUPPORTS_OPERATOR_DESERIALIZATION_WALKER (attribute access downstream); on older cores (including Astro Runtime 3.2 task SDK 1.2.x) the provider automatically dumps to a dict (subscript access). Check which shape arrives at runtime before choosing attribute vs dict access downstream, or set serialize_output=True to force the dict shape everywhere. The output_type class must be defined at module scope (nested classes cannot be deserialized from XCom).

@task.llm_branch

Only change: model= becomes llm_conn_id=.

@task.agent

This has the biggest API change. The Agent is no longer pre-built in user code.

Parameter mapping:

Key insight: Everything that was configured on the Agent() constructor now goes into either a top-level decorator parameter or agent_params. The agent_params dict is passed directly to pydantic-ai's Agent constructor. Prefer toolsets over agent_params["tools"]: the operator wraps each toolset in a LoggingToolset, so every tool call appears in the task log with timing.

enable_hitl_review behavior: the task generates a first draft, then blocks until a human acts. The reviewer uses the HITL Review tab/extra link on the task instance (chat UI from the provider's auto-registered hitl_review plugin) to request changes (agent regenerates with the feedback in its message history) or approve. Constraints: requires a string prompt, incompatible with durable=True, and the final (possibly regenerated) output is what flows to XCom. Warn users that the Dag run waits indefinitely at this task unless hitl_timeout is set. For headless testing, the plugin exposes REST endpoints under /hitl-review: GET /sessions/find, POST /sessions/feedback, POST /sessions/approve, POST /sessions/reject (query params dag_id, task_id, run_id, map_index).

@task.embed (NO EQUIVALENT — three replacement options)

The new provider does NOT include an embed decorator. Pick the replacement based on what the project needs:

Option A (recommended): LlamaIndexEmbeddingOperator (0.4.0, [llamaindex] extra). Connection-based, one task embeds the whole document list, and with persist_dir the resulting vector index is persisted for retrieval (pairs with LlamaIndexRetrievalOperator):

The operator returns {"chunks": [{"text", "metadata", "vector"}], ...}. Put a stable key into each document's metadata — it round-trips through chunking, so vectors can be mapped back to source records.

Option B: LlamaIndexHook for raw vectors (no operator, no persisted index). Shortest path when vectors go straight to a database:

Option C: plain @task with sentence-transformers (keeps the old local/offline behavior, no API cost; requires keeping sentence-transformers + torch in requirements):

Note on dimensions: switching from all-MiniLM-L6-v2 (384) to text-embedding-3-small (1536) changes vector size — existing stored embeddings must be regenerated, and fixed-size vector columns (e.g. pgvector vector(384)) need a schema change. Embed all texts in one task/batch call rather than .expand() per text: batching is one API round-trip and avoids per-task model loading.

Step 4: Update imports

The @task.llm, @task.agent, @task.llm_branch decorators are auto-registered by the provider. No explicit import needed beyond from airflow.sdk import task.

pydantic_ai imports for non-decorator usage (e.g., BinaryContent for multimodal) are still valid since the new provider depends on pydantic-ai-slim (>= 1.71.0 for provider 0.4.0).

Step 5: Update connections.yaml (if used for local testing)

For custom endpoints:

Step 6: Clean up env vars

The new provider reads model config from the pydanticai connection, so env vars that previously fed the model in code are usually redundant. Before removing any of them, grep the project (and any sibling scripts/services) to confirm nothing else still references them:

Candidates for removal only if no other code references them:

• OPENAI_API_KEY (now in the pydanticai connection's password field)
• OPENAI_BASE_URL (now in the connection's host field)
• Custom model name vars (now in the connection's extra.model)

If anything outside the migrated DAGs still uses them (other DAGs not yet migrated, helper scripts, non-Airflow services sharing the .env), leave them in place.

Keep AIRFLOW_CONN_* env vars for all connections.

Step 7: Verify

After migration, grep the codebase to confirm no stale references remain:

Verify:

• No imports from airflow_ai_sdk
• No Agent() objects created for @task.agent (unless used outside decorators)
• No model= parameter on LLM decorators (should be llm_conn_id=)
• All @task.embed replaced (LlamaIndex operator/hook or plain @task); stored embeddings regenerated if the model/dimensions changed
• Vision translation functions return [text, BinaryContent(...)] again if they were string-only-restricted under common-ai 0.1.x
• Downstream consumers of output_type=BaseModel results use the XCom shape that actually arrives (dict on older cores, instance on newer; serialize_output=True pins it)
• pydanticai connection configured in .env or connections.yaml
• requirements.txt has apache-airflow-providers-common-ai[...] instead of airflow-ai-sdk[...]; torch/sentence-transformers removed if no longer used
• Run the Dags end-to-end: tasks with enable_hitl_review=True or require_approval=True wait for human input, so the test plan must include acting on them (UI tab or /hitl-review REST)

Quick reference: New features in common-ai provider

These features are available after migration but have no airflow-ai-sdk equivalent: