A2A INTEGRATION · CUSTOM TOOL

Use GitDealFlow A2A with LangChain

The most popular Python and JS agent framework. Builders chaining multiple tools across LLM providers.

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

LangChain has experimental MCP support via langchain-mcp-adapters, and our MCP server works out of the box with that path. For A2A specifically, the cleanest integration is a small Tool subclass that wraps the JSON-RPC call. Both paths return the same data — pick MCP if you already use it elsewhere, A2A if you want a single HTTP dependency.

Install

pip install langchain langchain-core
# or
pip install langchain-mcp-adapters  # if you prefer MCP

Custom A2A Tool (Python)

from langchain_core.tools import tool
import requests

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

@tool
def gitdealflow_a2a(skill: str, args: dict | None = None) -> dict:
    """Call the GitDealFlow A2A agent.

    Args:
        skill: One of get_trending_startups, search_startups_by_sector,
               get_startup_signal, get_signals_summary, get_methodology.
        args: Optional dict of skill-specific arguments.
    """
    body = {
        "jsonrpc": "2.0", "id": 1,
        "method": "message/send",
        "params": {
            "message": {
                "role": "user",
                "parts": [{"kind": "data", "data": {"skill": skill, "args": args or {}}}],
            }
        },
    }
    return requests.post(A2A_URL, json=body, timeout=15).json()

MCP via langchain-mcp-adapters

# pip install langchain-mcp-adapters
from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent

client = MultiServerMCPClient({
    "gitdealflow": {
        "command": "npx",
        "args": ["@gitdealflow/mcp-signal@latest"],
        "transport": "stdio",
    },
})

tools = await client.get_tools()
agent = create_react_agent("openai:gpt-4o-mini", tools)
response = await agent.ainvoke(
    {"messages": [{"role": "user", "content": "What's trending in fintech?"}]}
)

What you can ask

agent.invoke({'messages': [...top 20 trending...]})
Chain: GitDealFlow lookup → SerpAPI for press → Crunchbase API for funding → memo writer
Filter: only return startups in user's geography
Compose: weekly digest from get_trending_startups + LLM summarization

Gotchas

If you use `create_react_agent`, the model occasionally calls the tool with malformed args. Validate before passing through.
The A2A endpoint has no rate limit at the application layer, but if you call it 100x/sec from a batch loop, the upstream CDN may briefly throttle. Add a 100ms sleep between calls in tight loops.

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

Custom A2A Tool (Python)

from langchain_core.tools import tool import requests A2A_URL = "https://signals.gitdealflow.com/api/a2a" @tool def gitdealflow_a2a(skill: str, args: dict | None = None) -> dict: """Call the GitDealFlow A2A agent. Args: skill: One of get_trending_startups, search_startups_by_sector, get_startup_signal, get_signals_summary, get_methodology. args: Optional dict of skill-specific arguments. """ body = { "jsonrpc": "2.0", "id": 1, "method": "message/send", "params": { "message": { "role": "user", "parts": [{"kind": "data", "data": {"skill": skill, "args": args or {}}}], } }, } return requests.post(A2A_URL, json=body, timeout=15).json()

MCP via langchain-mcp-adapters

# pip install langchain-mcp-adapters from langchain_mcp_adapters.client import MultiServerMCPClient from langgraph.prebuilt import create_react_agent client = MultiServerMCPClient({ "gitdealflow": { "command": "npx", "args": ["@gitdealflow/mcp-signal@latest"], "transport": "stdio", }, }) tools = await client.get_tools() agent = create_react_agent("openai:gpt-4o-mini", tools) response = await agent.ainvoke( {"messages": [{"role": "user", "content": "What's trending in fintech?"}]} )

Gotchas

If you use `create_react_agent`, the model occasionally calls the tool with malformed args. Validate before passing through.

The A2A endpoint has no rate limit at the application layer, but if you call it 100x/sec from a batch loop, the upstream CDN may briefly throttle. Add a 100ms sleep between calls in tight loops.