pydantic-ai-firestore-memory

A Google Firestore MemoryStore for the PydanticAI harness Memory capability — the per-user Markdown notebook the harness injects before every model call and lets the model write through write_memory, on a shared, durable collection.

pip install pydantic-ai-firestore-memory

from pydantic_ai import Agent
from pydantic_ai_harness.memory import Memory
from pydantic_ai_firestore_memory import FirestoreMemoryStore

store = FirestoreMemoryStore(project_id="my-project", collection="agent_memory")
agent = Agent("anthropic:claude-sonnet-5", deps_type=Deps,
              capabilities=[Memory(store=store, namespace=lambda ctx: ctx.deps.user_id)])

Authentication uses Application Default Credentials (gcloud auth application-default login).

How it works

• One collection: a document per memory file (kind: "file", path, content, version, operation_id) and per operation receipt (kind: "op"). No index setup needed.
• Compare-and-swap: create, replace and delete each run in a Firestore transaction that reads the document, checks the harness version, then writes; Firestore aborts and retries on contention, so a stale version raises MemoryConflictError and never overwrites.
• Operation receipts are reserved transactionally, completed after the mutation, dropped if it fails — durable-execution replays return the original result; a reused id with different arguments raises MemoryOperationConflictError.
• search is the harness's bounded lexical search (SearchableMemoryStore); list_paths is a range query on path.

Docs: https://skamalj.github.io/agentstate-reducer/ · part of pydantic-ai-memory

License

MIT