A2A INTEGRATION · CUSTOM TOOL

Use GitDealFlow A2A with Pydantic AI

Type-safe Python agent framework. Python builders who want strict typing on every agent input and output.

Crunchbase API: $20K/yr. GitDealFlow A2A: free, no signup.

Pydantic AI's Tool decorator enforces typed args and return shapes. Wrap our A2A endpoint as a single Tool that takes a skill enum and a typed args dict, and the agent gets compile-time checking on every call. The MCP server also works via pydantic-ai's experimental MCP support if you prefer.

Install

pip install pydantic-ai

Typed Tool

from pydantic_ai import Agent, Tool, RunContext
from pydantic import BaseModel
from typing import Literal
import requests

A2A_URL = "https://signals.gitdealflow.com/api/a2a"

Skill = Literal[
    "get_trending_startups",
    "search_startups_by_sector",
    "get_startup_signal",
    "get_signals_summary",
    "get_methodology",
]

class A2AArgs(BaseModel):
    skill: Skill
    args: dict | None = None

async def gitdealflow_query(ctx: RunContext, params: A2AArgs) -> dict:
    """Call GitDealFlow A2A for live VC engineering signals."""
    body = {
        "jsonrpc": "2.0", "id": 1,
        "method": "message/send",
        "params": {"message": {"role": "user", "parts": [
            {"kind": "data", "data": {"skill": params.skill, "args": params.args or {}}},
        ]}},
    }
    return requests.post(A2A_URL, json=body, timeout=15).json()

agent = Agent(
    "openai:gpt-4o-mini",
    tools=[Tool(gitdealflow_query)],
    system_prompt="Use gitdealflow_query for live engineering signals.",
)

What you can ask

result = await agent.run('Who is trending in fintech?')
Compose with other Pydantic AI tools (web search, vector store) for an end-to-end deal-memo flow.
Stream tokens with agent.run_stream() while the tool fetches in the background.
Validate output with a result_type=DealMemo Pydantic model.

Gotchas

Pydantic AI revalidates tool args on every call — keep your A2AArgs model lean to avoid latency.
If you want the agent to use multiple skills in one turn, raise tool_call_limit (default 10) in Agent config.

References

Try it without setup

The interactive playground lets you send live JSON-RPC requests against the A2A endpoint with no install, no auth. Pick a skill, hit send, see the response.

Full launch story: I made my VC deal flow callable by Claude.

Other framework integrations

Typed Tool

from pydantic_ai import Agent, Tool, RunContext from pydantic import BaseModel from typing import Literal import requests A2A_URL = "https://signals.gitdealflow.com/api/a2a" Skill = Literal[ "get_trending_startups", "search_startups_by_sector", "get_startup_signal", "get_signals_summary", "get_methodology", ] class A2AArgs(BaseModel): skill: Skill args: dict | None = None async def gitdealflow_query(ctx: RunContext, params: A2AArgs) -> dict: """Call GitDealFlow A2A for live VC engineering signals.""" body = { "jsonrpc": "2.0", "id": 1, "method": "message/send", "params": {"message": {"role": "user", "parts": [ {"kind": "data", "data": {"skill": params.skill, "args": params.args or {}}}, ]}}, } return requests.post(A2A_URL, json=body, timeout=15).json() agent = Agent( "openai:gpt-4o-mini", tools=[Tool(gitdealflow_query)], system_prompt="Use gitdealflow_query for live engineering signals.", )