For People In A Hurry: How to Set Up Your Own Deep Local Research Agent

README.md

Note

This is a quick copy-paste-observe guide for people in a hurry to quickly set up a Deep Local Research Agent where there are no reliance on proprietary web search api usage while still being useful in providing detailed reports on selected user topics.

For this guide, we will be using LearningCircuit's local_deep_research implementation with the following setup structure:

• [Web Search] - Dockerised
• [Deep Local Research] - Dockerized
• [Ollama] - Host

Instructions 📘

1. Install Ollama using the command below:

curl -fsSL https://ollama.com/install.sh | sh

2. Download ollama LLM model by using the command below:

ollama pull <model_name>
#Eg. ollama pull qwen3:8b

3. Install local-deep-research using the command below:

cd $HOME
git clone https://github.com/LearningCircuit/local-deep-research.git --depth 1 --single-branch --branch main && cd local-deep-research

4. Create a docker-compose.personal.yml with the contents below:

services:
  local-deep-research:
    container_name: ldr_ldr_c
    image: localdeepresearch/local-deep-research:latest
    restart: unless-stopped
    network_mode: host
    volumes:
      - deep-research:/data
    environment:
      LDR_DATA_DIR: /data
  searxng:
    container_name: ldr_searxng_c
    image: searxng/searxng:latest
    restart: unless-stopped
    ports:
      - "8081:8080"
    volumes:
      - ./searxng_data/settings.yml:/etc/searxng/settings.yml

volumes:
  deep-research:

Note

You might wondering why create our own docker-compose yaml file when the repository already provides its own? This is to cut through the many unnecessary boilerplate logic in the original and provide a working reference.

5. Create searxng_data directory using the command below:

mkdir searxng_data && cd searxng_data

6. Create settings.yml in directory, searxng_data:

use_default_settings: true

general:
  debug: true

server:
  bind_address: "0.0.0.0"
  port: 8080
  limiter: false  # 🔥 disables rate limiting (important)

search:
  formats:
    - html
    - json

botdetection:
  enabled: false  # 🔥 disables bot blocking

engines:
  - name: karmasearch
    disabled: true

Note

Note that we are disabling karmasearch as it easily hits its limit when utilised by the Deep Local Research agent. Once reached, it halts the entire research workflow.

7. Run the docker containers using the command below:

docker compose -f docker-compose.personal.yml up -d

8. Access the dashboard at http://0.0.0.0:5000/auth/login.

References

1. LearningCircuit's Deep Local Research - https://github.com/LearningCircuit/local-deep-research
2. Ollama - https://ollama.com/