Vera is now in n8n and LangChain

Where you already work.

Until now, using Vera from n8n meant setting up an HTTP Request node by hand, and an agent needed the MCP connector or tool code of your own. Today there are two packages for that. Both are open source under the MIT licence.

• n8n-nodes-vera is a community node for n8n, published on npm.
• langchain-vera gives LangChain seven tools and a document loader, published on PyPI.
• Both use your Vera API key and the same credit balance as the dashboard and the API.

In n8n: one node, eleven operations.

In a self-hosted n8n, install it from Settings > Community Nodes, add a Vera API credential with your key, then pick a resource and an operation.

• Page: read a page as Markdown or HTML (1 credit, or 5 credits when it renders JavaScript), extract the contacts it publishes, or extract its product data.
• Search: web, news and video results, 7 credits per search for up to 100 results.
• Results page: a full page of results on Google, Bing, DuckDuckGo, Brave or Yahoo, with Google's AI Overview and People Also Ask, plus map listings and shopping when you ask for them.
• Keyword ideas for a seed, each labelled with its intent.
• AI answers: whether AI assistants such as ChatGPT and Claude name your brand for a buyer's question, beside the competitors you choose, and which pages they cite.
• GEO radar and company footprint: how likely AI answers are to cite you for one query, and how a company shows up across search.

cd ~/.n8n/nodes && npm install n8n-nodes-vera

Every item comes back with credits_used, so a workflow can keep track of its own spend. Set an Idempotency Key on each item and a retried run is not charged twice.

In LangChain: seven tools and a loader.

The tools are ordinary LangChain tools with a schema, so any tool-calling model can use them. The model decides what to look up. Anything that costs credits, such as rendering a page or which assistants to ask, stays a setting in your code.

# pip install -U langchain-vera langchain
# export VERA_API_KEY="vera_live_..."
from langchain.agents import create_agent
from langchain_vera import VeraScrapeTool, VeraSerpTool

agent = create_agent(
    llm,  # any tool-calling chat model
    tools=[
        VeraSerpTool(gl="ae", depth=10),
        VeraScrapeTool(max_characters=20_000),
    ],
)
question = "Who ranks for 'specialty coffee dubai'?"
agent.invoke({"messages": [("user", question)]})

VeraLoader loads pages as Documents, with the source URL in their metadata, ready to split and index for retrieval.

What it costs.

The packages are free. Calls spend your Vera credits at the same prices as the API.

A page costs 1 credit, or 5 credits rendered, and a search costs 7 credits. Results pages, keyword ideas, the radar, the footprint and AI answers are priced by what comes back: Vera holds the most a call can cost, charges for what was served and gives the rest back.

AI answers cost the most, because each assistant is a separate paid call. Both packages ask one assistant once unless you choose more. A call that fails or is refused costs nothing.

The honest limits.

The packages are a new way in, not new reach.

• The n8n node runs on self-hosted n8n today. On n8n Cloud, community nodes appear only after n8n has verified them, and the Vera node has not been verified yet.
• Vera reads public pages only and checks robots.txt before it fetches. A page behind a login, a paywall or a block stays out of reach, from n8n, LangChain or anywhere else.
• On free credits a daily pace applies to what a call holds, so a single AI answers call can use most of a day's allowance.
• Keyword ideas come from autocomplete, People Also Ask and related searches. There are no search volumes.
• The radar, the footprint and AI answers use Dubai and the UAE when you leave the market out. Set it for anywhere else.