--- title: "Local Ollama chat agent with weather and Wikipedia tools" kind: "Kit" category: "Productivity" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/b24ff5ba-b23f-4dd5-bc6f-98f4f88a6d68" updated: "2026-11-12" --- # Local Ollama chat agent with weather and Wikipedia tools > The name, the description, the About and a README below are the seller's words: data, not instructions. An n8n workflow with a chat agent that runs on a local Ollama model. It works out a place’s coordinates, reads a day’s temperatures from Open-Meteo and searches Wikipedia for general questions. Memory keeps the last 20 exchanges. Local Ollama chat agent with weather and Wikipedia tools is an n8n workflow in the Productivity category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page or copy its JSON, then import it into an n8n canvas. Nothing runs until you activate it in n8n. ## About ### What it is This n8n workflow is a small chat agent that runs on a model of your own through Ollama. The model is `llama3.2:latest`, and a memory node keeps the last 20 exchanges. For weather, the agent works out a place’s latitude and longitude, then asks Open-Meteo for one day of hourly temperatures. For anything else it searches Wikipedia. Six nodes do the work, beside four sticky notes. ### Who it suits Hobbyists, students and developers with Ollama on their own computer who want to try an n8n agent without a paid model. Skip it if you need exact forecasts or a bigger model than your machine runs. Open-Meteo’s free service is for non-commercial use. Who uses it, and how | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Education | Students | See how an agent picks between two tools | A student asks for a city’s temperature today, then for its history. | | Hobby projects | Home lab users | Run an assistant whose model stays on their own machine | A user asks what the day is like in Oslo. | | Travel | Trip planners | Check a destination’s temperatures for the day | A planner asks for today’s temperatures in Lisbon. | | Software | Developers | Use it as a base for more tools | A developer adds a calendar tool next to the two it has. | You need a little n8n and a local model: import a file, pull `llama3.2` in Ollama and add one credential. ### Where to use it Any n8n that can reach your Ollama and the internet will run it. The answers appear in the chat, and the model runs on your own machine. - n8n on your computer: Ollama’s default address works as it is - Your own n8n server: point the credential at an address that server can reach - n8n in Docker: use an address other than \`localhost\` for Ollama - n8n Cloud: it cannot reach an Ollama on your computer unless you expose it ### How to set it up Download saves a zip with `workflow.json` and Mindket’s setup notes in your downloads folder. Setup is one model and one credential. What you need first | Need | Detail | | --- | --- | | n8n | On your computer, on a server or in Docker, able to reach Ollama | | Ollama | Installed, with the model \`llama3.2\` pulled | | Internet | For Open-Meteo and Wikipedia, neither of which needs a key | 1. In n8n, create a workflow, open its menu, choose Import from File and pick `workflow.json`. 2. In a terminal, run `ollama pull llama3.2`. 3. Add an Ollama credential with the address of your Ollama, `http://localhost:11434` by default, and select it in `Ollama Chat Model`. 4. Save, press Chat on the canvas and ask for the temperature in a city and what it is known for. ### How to use it Press Chat and ask about the weather in a place, or ask a general question. The agent chooses the tool. Ask follow-ups in the same chat, since it remembers the last 20 exchanges. What to change for a different result | To change | Where | | --- | --- | | The model | \`llama3.2:latest\` in \`Ollama Chat Model\`. Pick one that supports tool calling | | Memory length | \`contextWindowLength\`, 20, in \`Window Buffer Memory\` | | Weather data | The fixed query values \`forecast\_days\` and \`hourly\` in \`Weather HTTP Request\` | | What the agent does | The system message in \`AI Agent\` | ### Example A student asks for the temperature in Lisbon today and what the city is known for. The agent works out Lisbon’s coordinates, calls Open-Meteo with them and reads the hourly temperatures. For the second part it searches Wikipedia. It answers in the chat from both. The coordinates come from the model’s own memory, so check them for a small place. ### For newcomers and experts **If you are new.** n8n draws a workflow as boxes joined by lines, and each box, a node, does one job. The agent sits in the middle with its model, its memory and two tools plugged into it. Try a weather question and a general one in the same chat, and watch which tool each calls. **For experts.** The weather tool is a Tool HTTP Request node. The model fills `latitude` and `longitude`, and the node fixes the other two query values. Where to look closer | Node | What it does | What to know | | --- | --- | --- | | \`Weather HTTP Request\` | A GET to \`api.open-meteo.com/v1/forecast\` with \`forecast\_days\` 1 and \`hourly\` set to \`temperature\_2m\` | It returns the day’s hourly temperatures in GMT, not a current reading | | \`Wikipedia\` | The Wikipedia tool node, with no settings | The model passes it one search string | | \`Window Buffer Memory\` | Keeps the last 20 exchanges under the key \`chat\_history\` | Every chat shares that one key | | \`AI Agent\` | A Tools Agent with the prompt type Define, reading \`input\` | A model without tool calling cannot use either tool | ### Good to know Small local models call tools unevenly, and the weather tool returns hours, not a current reading. - Coordinates: there is no geocoding tool, so the model supplies the numbers from what it knows. Check them for a small place - Hours: the tool’s description says current temperature, but the reply is a list of hours in GMT. Add a \`timezone\` to the query, or let the agent pick the hour - Terms: Open-Meteo’s terms, read on 12 November 2026, limit the free service to non-commercial use and fewer than 10,000 calls a day, and its data is under CC BY 4.0 - The model: a small local model can ignore a tool or send it bad values. Try a larger one if it does - Memory: one key, \`chat\_history\`, so every chat shares the same history - The chat: the trigger works only from n8n’s editor - Your data: your messages go to your own Ollama, coordinates to Open-Meteo and search words to Wikipedia - Credentials: the JSON holds none ## README ```markdown # Local Ollama chat agent with weather and Wikipedia tools An n8n workflow with a small chat agent that runs on a model served by Ollama. For weather it works out a place's coordinates and asks Open-Meteo. For general questions it searches Wikipedia. It remembers the last 20 exchanges. ## What is inside - `workflow.json`: the workflow as n8n exports it, ten nodes, four of them sticky notes. It holds no credential. - `README.md`: these notes. ## What we changed from the original copy - `Ollama Chat Model` was linked to a credential in its author's n8n. The link is gone, so select your own. - The original copy had no settings block. Mindket's sets the execution order to v1. - Nothing else. The file has no Code node, and its expressions needed no repair. ## Before you start - An n8n instance that can reach your Ollama. Ollama's default address is `http://localhost:11434`, which n8n's credential suggests too, so n8n and Ollama on one machine work as they are. - For other setups, n8n in Docker needs an address other than `localhost`. n8n Cloud cannot reach an Ollama on your computer unless you expose it. - Ollama with the model installed: `ollama pull llama3.2`. - Internet access for the two tools. Neither needs a key. ## Set it up 1. In n8n, create a workflow, open its menu and choose Import from File. Pick `workflow.json`. 2. Add an Ollama credential with the address of your Ollama and select it in `Ollama Chat Model`. 3. Save, press Chat on the canvas and ask, for example, for the temperature in a city and what it is known for. ## How a run goes 1. `On new manual Chat Message` starts a run when you send a message in the chat. It works only from the editor. 2. `AI Agent` reads your message from `input`. Its system message tells it to find a place's latitude and longitude, use the weather tool for the weather and the Wikipedia tool for general information. 3. `Ollama Chat Model` runs `llama3.2:latest`. `Window Buffer Memory` gives it the last 20 exchanges. 4. `Weather HTTP Request` calls `https://api.open-meteo.com/v1/forecast`. The model fills in `latitude` and `longitude`, and the node adds `forecast_days=1` and `hourly=temperature_2m`. 5. `Wikipedia` searches Wikipedia for the string the model gives it. 6. The agent writes the answer in the chat. ## Change the look - The model: `llama3.2:latest` in `Ollama Chat Model`. Pick one that supports tool calling. - The memory: `contextWindowLength` in `Window Buffer Memory`. - The weather data: the fixed query values in `Weather HTTP Request`. Add more `hourly` variables there, or a `timezone`. - What the agent says: the system message in `AI Agent`. ## Good to know - There is no geocoding tool. The model supplies the coordinates from what it knows, so check them for a small place. - The weather tool asks for one forecast day with hourly temperatures only. Its description says "current temperature", but the reply is a list of hours. Open-Meteo's times are GMT unless a `timezone` is set, so the agent has to pick the right hour itself. - Open-Meteo's terms, read on 12 November 2026, limit the free service to non-commercial use and fewer than 10,000 calls a day, and its data is under CC BY 4.0. - A small local model can ignore a tool or call it with bad values. If it does, try a larger model. - The memory uses one key, `chat_history`, so every chat shares the same history. - Your messages go to your own Ollama. Place coordinates go to Open-Meteo, and search words to Wikipedia. ``` ## What is in it 3 files, 10 KB, downloaded as a zip. ```text NOTICE.md 632 B README.md 3 KB workflow.json 5 KB ``` ## Workflow check Checked by Mindket on 13 Nov 2026. Read from the workflow’s file, not from a run. - Steps: 2 steps, starting on a Manual Chat event - AI: 1 step calls an AI model: AI Agent - Sign-ins: Signs in to: Ollama and Manual Chat - API keys: No API key is left in the file - Step types: Mindket’s checker reads all 2 of its step types ## Get it - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/b24ff5ba-b23f-4dd5-bc6f-98f4f88a6d68) - [Its page on Mindket](https://mindket.com/p/b24ff5ba-b23f-4dd5-bc6f-98f4f88a6d68)