Skip to content

djangorestframework-pydantic-ai

Expose djangorestframework-services services and selectors as a Pydantic-AI toolset, so a plain pydantic_ai.Agent can call them as tools — no MCP server and no AG-UI bridge in the path.

What this is

A single thin adapter, SpecToolset, that turns a name -> spec mapping into agent tools. Every call routes through drf-services' transport-neutral surface:

  • dispatch_spec executes the service or selector (validation, instance resolution, output-selector re-fetch);
  • build_offline_context supplies a synthetic request / view / principal so permission_classes, extend_queryset, and context providers keep working;
  • enforce_permissions runs spec.permission_classes — dispatch_spec deliberately does not, so building a toolset naively on it would skip authorization;
  • spec_to_json_schema derives each tool's parameter schema;
  • render_spec_output renders the result through the spec's serializer.

The result is that an agent driving these tools sees exactly what an HTTP client would: the same validation errors, the same permission_classes checks (both class-level has_permission and object-level has_object_permission on the resolved row), the same rendered payloads — without the network hop, and without standing up an MCP server.

How it compares

If you already run an MCP server, django-ag-ui's DrfMcpToolset exposes its tools to Pydantic-AI through the in-process MCP surface. SpecToolset is for the case where you want the specs as tools without any MCP layer: it depends only on djangorestframework-services and pydantic-ai-slim.

Next steps