LangChain + Govyn — Govern Your LangChain Agents

LangChain agents calling OpenAI can burn through your API budget in minutes with recursive tool calls. Without a governance layer, a single runaway chain can generate hundreds of completions before you notice — and there's no way to enforce spending limits or model restrictions at the framework level.

How it works

LangChain
Your agents
─ HTTPS →
Govyn Proxy
Policy · Budget · Logs
─ API →
OpenAI API
LLM provider

Step-by-step setup

1

Start the Govyn proxy

bash 
npx govyn start --config govyn.yaml
2

Point LangChain at Govyn

python 
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="gpt-4o",
    openai_api_base="http://localhost:4111/v1",
    openai_api_key="gvn_agent_langchain_01"
)
3

Run your chain as usual

python 
from langchain.agents import AgentExecutor, create_openai_tools_agent

agent = create_openai_tools_agent(llm, tools, prompt)
executor = AgentExecutor(agent=agent, tools=tools)
result = executor.invoke({"input": "Summarize Q4 revenue"})

Example policy

Define governance rules for your LangChain agents in a simple YAML file.

govyn.yaml 
agents:
  langchain_01:
    budget:
      daily: $5.00
      monthly: $100.00
    models:
      allow: [gpt-4o, gpt-4o-mini]
      deny: [gpt-4-32k]
    rate_limit:
      requests_per_minute: 30
    logging:
      replay: true
      redact_pii: true

Why use Govyn with LangChain?

Per-agent daily and monthly budget caps
Model allowlists — block expensive models
Full request/response replay for debugging chains
Rate limiting to prevent runaway tool loops
PII redaction in logged completions
Zero code changes — just swap the base URL

Get started in 5 minutes

Add governance to your LangChain agents with a single config change. No code rewrites.

Read the docs

Frequently asked questions

Do I need to change my LangChain code to use Govyn?
No. You only change the base URL and API key when constructing your ChatOpenAI instance. All LangChain chains, agents, and tools continue to work exactly as before — Govyn is transparent at the HTTP level.
Does Govyn add latency to LangChain agent calls?
Govyn adds sub-millisecond overhead per request. Policy evaluation happens in-memory and the proxy forwards to OpenAI in a single async hop. This is negligible compared to LLM inference time, especially for multi-step chains.
Can I set different budgets for different LangChain agents?
Yes. Each agent gets a unique Govyn API key with its own budget, model restrictions, and rate limits. You can run dozens of LangChain agents through a single proxy, each with independent governance rules.

Related integrations

CrewAI + OpenAI

Add budget controls and policy enforcement to CrewAI multi-agent crews using OpenAI. Govern every agent in your crew independently.
Python + Any Provider

Add governance to any Python AI agent. Works with requests, httpx, and the OpenAI SDK. Budget limits, policy enforcement, full replay.
LangChain + Anthropic

Govern LangChain agents using Claude models. Enforce budgets, restrict models, and log every completion with Govyn's proxy.

Explore more

FROM OUR BLOG
Proxy vs SDK: Why Architecture Matters for AI Agent Governance

SDK wrappers are door locks. Proxies are walls. A deep technical comparison of both governance architectures for AI agents in production.

FROM OUR BLOG
We Cut Our AI API Bill by 73% Without Changing a Single Line of Agent Code

How smart model routing through a proxy cut our OpenAI and Anthropic bill from $2,140/mo to $578/mo. Zero code changes. Just YAML.

POLICY TEMPLATE
Budget Control Policy

Set daily and monthly spending limits for AI agents. Prevent runaway costs with hard budget caps enforced at the proxy level.

POLICY TEMPLATE
Smart Model Routing Policy

Automatically route AI agent requests to cheaper models when possible. Cut LLM costs by 60-80% with smart model routing policies.