Skip to main content

ACP adapter for pydantic-ai agents.

Project description

pydantic-acp

pydantic-acp adapts pydantic_ai.Agent instances to the ACP agent interface.

Public Entry Points

  • run_acp(...)
  • create_acp_agent(...)
  • AdapterConfig
  • AcpSessionContext
  • AgentSource, AgentFactory, StaticAgentSource, FactoryAgentSource
  • AgentBridgeBuilder
  • provider protocols
  • capability bridges
  • hook projection and existing-hook introspection
  • host backends

Implemented Scope

  • Milestone 1-7 scope is implemented
  • the adapter includes session lifecycle, session-local model control, providers, approvals, bridges, projection maps, and host backends
  • slash commands are available for /model, /tools, /hooks, and /mcp-servers

Quick Start

from pydantic_acp import run_acp
from pydantic_ai import Agent

agent = Agent("test", name="demo-agent")
run_acp(agent=agent)

Factory-backed usage:

from pydantic_acp import AcpSessionContext, create_acp_agent
from pydantic_ai import Agent

def build_agent(session: AcpSessionContext) -> Agent[None, str]:
    return Agent("test", name=f"demo-{session.cwd.name}")

acp_agent = create_acp_agent(agent_factory=build_agent)

File-backed session persistence:

from pathlib import Path

from pydantic_acp import AdapterConfig, FileSessionStore, run_acp

run_acp(
    agent=agent,
    config=AdapterConfig(
        session_store=FileSessionStore(base_dir=Path(".acp-sessions")),
    ),
)

Projection map example:

from pydantic_acp import FileSystemProjectionMap, run_acp

run_acp(
    agent=agent,
    projection_maps=(
        FileSystemProjectionMap(
            default_read_tool="read_file",
            default_write_tool="write_file",
            default_bash_tool="execute",
        ),
    ),
)

Hook projection example:

from pydantic_acp import HookProjectionMap, run_acp

run_acp(
    agent=agent,
    projection_maps=(
        HookProjectionMap(
            hidden_event_ids=frozenset({"after_model_request"}),
        )
    ),
)

Major Features

  • ACP session lifecycle support
  • transcript and model-message history replay
  • generic tool projection with ACP tool updates
  • projection maps for filesystem diffs and bash command previews
  • session-local models, modes, config options, and plan updates
  • deferred approval bridging
  • provider seams for host-owned state
  • capability bridges for hooks, history processors, prepare-tools, and MCP
  • hook introspection and HookProjectionMap rendering for existing Hooks capabilities
  • session-scoped host backends

Examples

See examples/pydantic/ for focused SDK examples covering static agents, factories, providers, bridges, approvals, host-context usage, and hook projection.

Key entry points:

  • examples/pydantic/hook_projection.py
  • examples/pydantic/my_agent.py

For fuller workspace documentation, see the root README.md and the docs/ directory.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pydantic_acp-0.3.1.tar.gz (35.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

pydantic_acp-0.3.1-py3-none-any.whl (49.6 kB view details)

Uploaded Python 3

File details

Details for the file pydantic_acp-0.3.1.tar.gz.

File metadata

  • Download URL: pydantic_acp-0.3.1.tar.gz
  • Upload date:
  • Size: 35.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.7

File hashes

Hashes for pydantic_acp-0.3.1.tar.gz
Algorithm Hash digest
SHA256 62d84cc4ca1b931a855ed991713bf3daf50ce284297284c671158cc507ed306a
MD5 3250d53004ed2f3dffa180503919f5da
BLAKE2b-256 2bb7c907983af7fc1bb5595a011a37b719a155186369e76404be6e6b56594fa8

See more details on using hashes here.

File details

Details for the file pydantic_acp-0.3.1-py3-none-any.whl.

File metadata

  • Download URL: pydantic_acp-0.3.1-py3-none-any.whl
  • Upload date:
  • Size: 49.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.7

File hashes

Hashes for pydantic_acp-0.3.1-py3-none-any.whl
Algorithm Hash digest
SHA256 fd2632aa62a3aad448150a3950e2b5935d25e4a7441458e8e38e5ccc0982db73
MD5 2f68140d7047f6a39c22c27bf8171b3a
BLAKE2b-256 5ce0a6be4d34a7414809e9356f924d6d88a39954fa4fec239db9200fe89548f5

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page