DEV Community

Cover image for 181 more tools for your LangGraph agent — in two lines
Alex
Alex

Posted on Originally published at modeldev.modelmarket.dev

181 more tools for your LangGraph agent — in two lines

#python #langchain #opensource #showdev

181 more tools for your LangGraph agent — two lines.

You already have a LangGraph (or CrewAI, or AutoGen) agent.

You do not want to rebuild randomness, delay proofs, consensus, or transport math as homemade tools.

Two lines. Live Hub capabilities become native framework tools — with a hard budget ceiling and signed receipts.

from aimarket_bridges.langchain import aimarket_tools

tools = aimarket_tools("https://modelmarket.dev", intent="verifiable randomness")
Enter fullscreen mode Exit fullscreen mode

Counted 2026-09-12 from the live Hub manifest: 181 federated capabilities across 14 hubs. The old "47" in the README is stale — that was one peer's catalogue, not the whole pier.

Crypto stays optional. Free trial exists so you can try before any wallet (5 invokes per visitor per hour on the public Hub).

aimarket-bridges landing — two lines into the live Hub

Install (one extra — frameworks disagree on pydantic) 

pip install "aimarket-bridges[langgraph]"
# or: pip install "aimarket-bridges[crewai]"
# or: pip install "aimarket-bridges[autogen]"
Enter fullscreen mode Exit fullscreen mode

Core has no framework dependency. Install only the adapter you use. That's what PyPI 0.1.0 ships.

Framework Import Returns
LangChain / LangGraph aimarket_bridges.langchain list[StructuredTool]
CrewAI aimarket_bridges.crewai list[crewai.tools.BaseTool]
AutoGen aimarket_bridges.autogen list[autogen_core.tools.BaseTool]

Same signature everywhere:

aimarket_tools(
    "https://modelmarket.dev",
    intent="verifiable randomness",  # rank the catalogue
    limit=12,                        # don't dump 181 tools into context
    max_price_usd=0.05,              # filter at build time
    budget_usd=1.0,                  # hard ceiling across every call
)
Enter fullscreen mode Exit fullscreen mode

The two-liner — native StructuredTool, not a stub

Minimal LangGraph sketch 

from langgraph.prebuilt import create_react_agent
from aimarket_bridges.langchain import aimarket_tools

tools = aimarket_tools(
    "https://modelmarket.dev",
    intent="verifiable randomness",
    limit=8,
    budget_usd=0.50,
)

# wire `tools` into your graph / create_react_agent as usual
agent = create_react_agent(model, tools)
Enter fullscreen mode Exit fullscreen mode

What you get back are real Hub capabilities (randomness, consensus, delay, transport, …) — not stubs. Each call can return a signed receipt kept out of the tool text (so the model doesn't eat a blob). Verify against the capability's origin key, not the hub's — the catalogue is federated.

Refusals come back as readable sentences. The graph keeps moving. Transport failures still raise — those the model can't fix.

Why a budget ceiling is the point

max_price_usd filters at build. budget_usd stops a looping agent.

Once a tool is in the agent's registry, the agent decides when to call it.

max_price_usd filters at build time. budget_usd is a hard stop before each call — a looping agent cannot outspend the cap.

That's the same frugality story we care about on the demand side — without asking you to abandon LangGraph.

Without a framework 

from aimarket_bridges import fetch_catalog, HubClient

caps = fetch_catalog("https://modelmarket.dev", intent="consensus")
with HubClient("https://modelmarket.dev", budget_usd=0.50) as hub:
    result = hub.invoke(caps[0], {"values": [1.0, 2.0, 3.0, 100.0]})
    print(result.output, result.receipt_verified)
Enter fullscreen mode Exit fullscreen mode

Soft star the pier

If you ship agents on LangGraph / CrewAI / AutoGen and want Hub tools without rewriting your stack:

⭐ github.com/alexar76/aimarket-bridges

📦 pip install "aimarket-bridges[langgraph]"

🌐 bridges landing · guide

MIT/Apache ecosystem. Self-hosted Hub if you want your own pier later — start with the two-liner.

Top comments (0)