Skip to content

About

Task Planning and Tracking toolset for Pydantic AI agents, enabling hierarchical task management with subtasks, PostgreSQL storage for multi-tenancy, and an event system for webhooks and callbacks.

Topics

Resources

Contributing

Stars

76 stars

Watchers

3 watching

Forks

Repository files navigation

Todo Toolset for Pydantic AI

Task Planning and Tracking for AI Agents

PyPI version Python 3.10+ License: MIT CI Pydantic AI

Capabilities API — plug-and-play with one line  •  Subtasks & Dependencies — hierarchical task management  •  PostgreSQL Storage — persistent multi-tenant tasks  •  Event System — webhooks and callbacks


⬆️ Upstreamed to pydantic-ai-harness

Working together with the Pydantic team, we folded this library into the official pydantic-ai-harness — it now lives in pydantic_ai_harness/planning (PR #404, merged), which supersedes this package.

For new projects, use the harness — it is maintained by Pydantic alongside Pydantic AI itself. This repository stays on PyPI and keeps working for everyone already depending on it.

from pydantic_ai import Agent
from pydantic_ai_harness.planning import Planning

agent = Agent("openai:gpt-5.4", capabilities=[Planning(enable_subtasks=True)])

Tools were renamed

Here In the harness
write_todos write_plan
read_todos read_plan
add_todo add_task
update_todo_status / update_todo_statuses update_task_status / update_task_statuses
remove_todo remove_task
add_subtask, set_dependency, get_available_tasks unchanged

API mapping

Here In pydantic_ai_harness.planning
TodoCapability Planning
TodoStorage, AsyncMemoryStorage InMemoryPlanStore
AsyncPostgresStorage / create_storage("postgres", connection_string=…) PostgresPlanStore — takes a caller-owned asyncpg-compatible pool, so the harness pulls in no database driver
create_todo_toolset() PlanningToolset
get_todo_system_prompt() not needed — Planning surfaces the plan as an ephemeral tail reminder behind a CachePoint, so the cached prefix stays byte-identical across turns
Todo, TodoItem PlanItem
TodoEvent, TodoEventType, TodoEventEmitter PlanEvent, PlanEventType, PlanEventEmitter

The harness also adds SqlitePlanStore and RedisPlanStore, and a blocked task status, none of which exist here.

Todo Toolset for Pydantic AI adds task planning capabilities to any Pydantic AI agent. Your agent can create, track, and complete tasks with full support for subtasks, dependencies, and persistent storage.

Full framework? Check out Pydantic Deep Agents — complete agent framework with planning, filesystem, subagents, and skills.

Quick Start

The recommended way to add todo support is via the Capabilities API — one import, one line:

from pydantic_ai import Agent
from pydantic_ai_todo import TodoCapability

agent = Agent("openai:gpt-4.1", capabilities=[TodoCapability()])
result = await agent.run("Create a todo list for building a REST API")

TodoCapability automatically:

  • Registers all todo tools (add_todo, read_todos, write_todos, update_todo_status, remove_todo)
  • Injects a system prompt section describing the todo workflow (static by default, so todo updates never invalidate the provider's prompt cache; pass include_current_todos=True to embed the live task list instead)
  • Creates in-memory storage (or use your own)

With Storage Access

from pydantic_ai_todo import TodoCapability, TodoStorage

storage = TodoStorage()
agent = Agent("openai:gpt-4.1", capabilities=[TodoCapability(storage=storage)])

result = await agent.run("Plan a blog application")

# Access todos directly
for todo in storage.todos:
    print(f"[{todo.status}] {todo.content}")

With Subtasks and Dependencies

agent = Agent(
    "openai:gpt-4.1",
    capabilities=[TodoCapability(enable_subtasks=True)],
)

Enables add_subtask, set_dependency, and get_available_tasks tools with automatic cycle detection.

YAML Agent Definition

model: openai:gpt-4.1
instructions: "You are a project planner."
capabilities:
  - TodoCapability:
      enable_subtasks: true
agent = Agent.from_file("agent.yaml")

Installation

pip install pydantic-ai-todo

Or with uv:

uv add pydantic-ai-todo

Alternative: Toolset API

If you prefer the lower-level toolset approach (without capabilities):

from pydantic_ai import Agent
from pydantic_ai_todo import create_todo_toolset, get_todo_system_prompt, TodoStorage

storage = TodoStorage()
toolset = create_todo_toolset(storage=storage)

agent = Agent(
    "openai:gpt-4.1",
    toolsets=[toolset],
    system_prompt=get_todo_system_prompt(storage),
)

Note: With the toolset API, you need to wire get_todo_system_prompt() manually. TodoCapability handles this automatically.

Available Tools

Tool Description
read_todos List all tasks (supports hierarchical view)
write_todos Bulk write/update tasks
add_todo Add a single task
update_todo_status Update task status by ID
remove_todo Delete task by ID
add_subtask* Create child task
set_dependency* Link tasks with dependency
get_available_tasks* List tasks ready to work on

*Available when enable_subtasks=True

Storage Backends

In-Memory (Default)

from pydantic_ai_todo import TodoCapability, TodoStorage

storage = TodoStorage()
agent = Agent("openai:gpt-4.1", capabilities=[TodoCapability(storage=storage)])

Async Memory

from pydantic_ai_todo import TodoCapability, AsyncMemoryStorage

storage = AsyncMemoryStorage()
agent = Agent("openai:gpt-4.1", capabilities=[TodoCapability(async_storage=storage)])

PostgreSQL

Requires the postgres extra: pip install 'pydantic-ai-todo[postgres]'

from pydantic_ai_todo import TodoCapability, create_storage

storage = create_storage(
    "postgres",
    connection_string="postgresql://user:pass@localhost/db",
    session_id="user-123",  # Multi-tenancy
)
await storage.initialize()

agent = Agent("openai:gpt-4.1", capabilities=[TodoCapability(async_storage=storage)])

Event System

React to task changes:

from pydantic_ai_todo import TodoCapability, TodoEventEmitter, AsyncMemoryStorage

emitter = TodoEventEmitter()

@emitter.on_completed
async def notify_completed(event):
    print(f"Task done: {event.todo.content}")

@emitter.on_created
async def notify_created(event):
    print(f"New task: {event.todo.content}")

storage = AsyncMemoryStorage(event_emitter=emitter)
agent = Agent("openai:gpt-4.1", capabilities=[TodoCapability(async_storage=storage)])

API Reference

Capability

Class Description
TodoCapability Pydantic AI capability — recommended way to add todo support

Factory Functions

Function Description
create_todo_toolset() Create standalone toolset (lower-level API)
create_storage(backend, **options) Factory for storage backends
get_todo_system_prompt() Generate system prompt with current todos

Models

Model Description
Todo Task with id, content, status, parent_id, depends_on
TodoItem Input model for write_todos
TodoEvent Event data with type, todo, timestamp
TodoEventType CREATED, UPDATED, STATUS_CHANGED, DELETED, COMPLETED

Storage Classes

Class Description
TodoStorage Sync in-memory storage
AsyncMemoryStorage Async in-memory with CRUD
AsyncPostgresStorage PostgreSQL with multi-tenancy
TodoEventEmitter Event emitter for callbacks

Related Projects

Package Description
Pydantic Deep Agents Full agent framework (uses this library)
pydantic-ai-backend File storage and Docker sandbox
subagents-pydantic-ai Multi-agent orchestration
summarization-pydantic-ai Context management
pydantic-ai The foundation — agent framework by Pydantic

Contributing

git clone https://github.com/vstorm-co/pydantic-ai-todo.git
cd pydantic-ai-todo
make install
make test  # 100% coverage required

License

MIT — see LICENSE


Need help implementing this in your company?

We're Vstorm — an Applied Agentic AI Engineering Consultancy
with 30+ production AI agent implementations.

Talk to us



Made with ❤️ by Vstorm

About

Task Planning and Tracking toolset for Pydantic AI agents, enabling hierarchical task management with subtasks, PostgreSQL storage for multi-tenancy, and an event system for webhooks and callbacks.

Topics

Resources

Contributing

Stars

76 stars

Watchers

3 watching

Forks

Releases

Contributors

Languages