MCP Mastery
About
Chapter 6
mid
~35 min

Tools and Function Calling — Stop pretending the model picks for you.

Stop pretending the model picks for you.

LangChain 0.3.x
LangGraph 0.2.x
Python 3.11
Reviewed 2026-05-16

Reading this chapter helps prevent 7 common LangChain mistakes.

The setup

You are here because tool contracts moved from notebook magic to product surface area. That is where the charming demo stops being charming and starts asking for state, tests, traces, and a budget.

This chapter gives tool schemas, bind_tools, execution loops, parallel calls, and error surfaces. The point is not to memorize one API call. The point is to know which abstraction deserves trust and which one is wearing a fake mustache.

Picture this

The production shape of tool contracts before the code starts freelancing.

Mental model

Think in contracts first. A LangChain runnable or LangGraph node is not a poetic suggestion; it is a boundary with inputs, outputs, config, callbacks, and failure modes. If you cannot describe those five things, you are not designing an agent system. You are hosting an improv night.

DecisionUse thisNot that
Primary movesmall typed tools with boring result envelopesletting the model discover your production API by vibes
EvidenceTraceable runs, typed payloads, and repeatable examplesA screenshot of one lucky response
Failure postureRetry, fallback, or interrupt with a reasonHope the model apologizes convincingly

Cocky, not careless

Confidence is earned by making the boring parts visible: schemas, state, traces, budgets, and tests. The model can be creative. Your architecture does not get that privilege.

Walkthrough

Start with a tiny runnable-shaped contract. Yes, it looks small. That is the point. Small contracts are how you stop a graph from becoming a haunted house.

python
from langchain_core.runnables import RunnableLambda


def describe_contract(payload: dict) -> dict:
    return {
        "input": payload["input"],
        "risk": "bounded",
        "next_step": "trace-and-test",
    }

chain = RunnableLambda(describe_contract).with_config(tags=["chapter-06"])
result = chain.invoke({"input": "ship the graph, not the vibes"})
print(result)

Now attach the concept to the actual chapter topic: tool schemas, bind_tools, execution loops, parallel calls, and error surfaces. The implementation pattern is deliberately boring: define the boundary, tag the run, record the outcome, then decide whether the next branch is cheap, risky, or worth human attention.

How tool contracts behaves once state, observability, and failure paths are admitted into the room.

Try this yourself

  1. Write a runnable or graph node for the smallest useful version of this chapter's pattern.
  2. Add tags and metadata before you run it. Future you deserves evidence, not folklore.
  3. Create one failure case on purpose and decide whether it should retry, fallback, interrupt, or stop.

Hall of Shame

Hall of shame: Returning a stack trace as a tool result and calling it a featurelangchain

This is the move that looks clever for ten minutes and becomes operational debt for six months. It hides the contract, loses the trace, and makes the next engineer debug vibes with a calendar invite. Beautiful work, if the goal was archaeology.

Why this matters in production

The boring checklist is undefeated:

  • Inputs are typed and validated.
  • Expensive branches have budgets.
  • Risky actions have approval gates.
  • Every meaningful run is traceable.
  • The failure mode is named before users name it for you.

You can now brag that...

  • You can explain tool contracts without hiding behind framework mysticism.
  • You know when to small typed tools with boring result envelopes.
  • You can spot letting the model discover your production API by vibes before it becomes a retro item.

References

Quiz

  1. What is the safest first move when designing tool contracts?

  2. Which signal says the implementation is ready for production review?

  3. Which anti-pattern does this chapter explicitly call out?