KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #codex #typescript #developer-tools #ai-agents #deepseek-harness #claude-code #ai-agent #cli #react #mcp

tourist

★ 27 Open GitHub ↗

Open-source, LLM-ready SERP and web scraping service

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Tourist🤳

[![PyPI version](https://badge.fury.io/py/tourist.svg)](https://badge.fury.io/py/tourist)

An open-source, low-cost, serverless application for SERP extraction and web scraping.

Work on LLM projects without worrying about credits, subscriptions, or rate-limits. Tourist is a free alternative to many mainstream SERP API services. Run Tourist on your machine or deploy it into your own AWS account.

> [!IMPORTANT]  
> Tourist is in early development. Features and API's may change unexpectedly.

## Overview
![tourist-architecture](https://raw.githubusercontent.com/pogzyb/tourist/main/docs/touristv020.png "Overview")

Tourist has both Service and Client components. The Service (HTTP API) handles requests from the Client (your app, agent, or scraper scripts). You're in control of both components! None of your data is ever processed or stored by third parties.

## Service

### Local deployment (for testing...)

> [!TIP]  
> Docker is recommended for running Tourist locally to handle dependencies for headless browsing.

1. `docker run -e X_API_KEY="whatever" -p 8000:8000 ghcr.io/pogzyb/tourist:latest app`

If the service came up correctly, you should see:
```
INFO:     Started server process [1]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)
```

Check the docs at `http://localhost:8000/docs`

### 🆕 MCP mode available with Azure deployment

1. `docker run -e X_API_KEY="whatever" -p 8000:8000 ghcr.io/pogzyb/tourist:latest mcp`

### AWS deployment

Prerequisites:
1. AWS Account and Credentials
2. S3 Bucket for managing an OpenTofu statefile
3. Docker

Read about the deployment container in the [deploy/](./deploy/DEPLOY.md) folder.

Example deployment command:

```
docker run \
    -v /var/run/docker.sock:/var/run/docker.sock \
    --env-file .env \
    ghcr.io/pogzyb/tourist-deploy:latest aws deploy \
    --state-bucket llmabda-statefile \
    --region us-east-1 \
    --mode app \
    --x-api-key SeCretTK3y
```

Use your endpoint: `https://xxxxxxxx.lambda-url.us-east-1.on.aws` (available the deployment outputs)

### 🆕 Azure deployment

Prerequisites:
1. Azure Account and Credentials (ARM_CLIENT_ID, ARM_CLIENT_SECRET, ARM_TENANT_ID, and ARM_SUBSCRIPTION_ID)
2. A resource group, storage account, and storage container for managing an OpenTofu statefile
3. Docker

```
docker run \
    -v /var/run/docker.sock:/var/run/docker.sock \
    --env-file .env \
    ghcr.io/pogzyb/tourist-deploy:latest azure deploy \
    --tofu-resource-group "tofu-resource-group" \
    --tofu-storage-account-name "tofustatemanagment" \
    --tofu-container-name "statefiles" \
    --x-api-key "MyS3creTkeey" \
    --mode mcp
```

Use your endpoint: `https://tourist.xxxxxxxx.eastus.azurecontainerapps.io` (available the deployment outputs)

> [!WARNING]  
> Tourist uses serverless infrastructure to keep costs extremely low; however these costs will not be $0.00 depending on how heavily you use your API.

> [!IMPORTANT]  
> Tourist uses the X-API-KEY authorization header to protect your API, you should set this value when deploying.

## Client

Build your own LLM tools, web scraping apps, or automated testing workflows with the Tourist client.

#### Python

You can use the python client to interact with your Tourist service.
Check out the [examples](https://github.com/pogzyb/tourist/tree/main/examples) folder for the complete code.

```
pip install tourist
```

##### LLM Tools

```
import asyncio

from llama_index.llms.ollama import Ollama
from llama_index.core.agent.workflow import FunctionAgent
from tourist.client import TouristScraper

# Assumes you're running locally,
# change this to your lambda endpoint if you've deployed to your aws account.
tourist_scraper = TouristScraper(
    func_urls="http://localhost:8000",
    x_api_key="whatever",
)

async def search_web(query: str):
    """Search places, items, people, or current events online."""
    serp_result = await tourist_scraper.aget_serp(
        query, search_engine="duckduckgo", max_results=3
    )
    return serp_result

async def main():
    llm = Ollama(model="ministral-3:3b", request_timeout=120.0)

agent = FunctionAgent(
        tools=[search_web],
        llm=llm,
        system_prompt="You are a helpful assistant that can search the web for information.",
        verbose=True,
    )

response = await agent.run(
        user_msg="Which country has the most medals in the 2026 winter olympics?"
    )
    print(str(response))

if __name__ == "__main__":
    asyncio.run(main())

```

tourist

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README