Skip to content

README.md

Latest commit

 

History

History
45 lines (32 loc) · 1.92 KB

README.md

File metadata and controls

45 lines (32 loc) · 1.92 KB

Browserless × LangChain (Python)

Runnable Python examples for the Browserless MCP server, using langchain-mcp-adapters + langgraph.

Setup

python -m venv .venv && source .venv/bin/activate
pip install -e .
pip freeze > requirements.txt
export BROWSERLESS_TOKEN=<your-token>
export ANTHROPIC_API_KEY=<your-key>

Get a Browserless token at account.browserless.io.

Notebooks

Notebook What it shows Tier
quickstart.ipynb Connect, list tools, call browserless_smartscraper directly, run a stateless ReAct agent 1 (stateless)
research_agent.ipynb Multi-step search → scrape → summarize using stateless tools 1 (stateless)
browser_agent.ipynb Multi-turn browser agent: navigate → snapshot → click → re-snapshot → extract 2 (stateful)

Note on Python session handling

langchain-mcp-adapters opens a fresh MCP ClientSession per tool.ainvoke() call by default — see the adapter README. That works for stateless tools, but for multi-turn browserless_agent flows you need to bind tools to a long-lived session:

async with client.session("browserless") as session:
    tools = await load_mcp_tools(session)
    # ... all agent calls go inside this block

Without it, each ainvoke() opens a fresh Mcp-Session-Id, the load balancer routes to a different browser, and state is lost between steps. browser_agent.ipynb shows the correct pattern.

The JS adapter does not have this requirement — see ../js/browser_agent.ts.

Run

jupyter nbconvert --to notebook --execute quickstart.ipynb --output quickstart.executed.ipynb
jupyter nbconvert --to notebook --execute research_agent.ipynb --output research_agent.executed.ipynb
jupyter nbconvert --to notebook --execute browser_agent.ipynb --output browser_agent.executed.ipynb