Skip to main content
Qdrant is the vector database that powers Docbot’s semantic search capabilities. This guide covers setting up Qdrant with Docker for local development.

​
Why Qdrant?

Docbot uses Qdrant because it’s:
  • Easy to run locally via Docker
  • Performant for document and code search
  • Well-suited for semantic similarity queries
  • Reliable for production use
Qdrant is currently required for Docbot. Support for other vector stores may be added in the future.

​
Docker setup

1

Install Docker

If you don’t have Docker installed:Verify installation:
docker --version
2

Pull the Qdrant image

Download the official Qdrant Docker image:
docker pull qdrant/qdrant
3

Start Qdrant container

Run Qdrant with persistent storage:
docker run -d \
  --name docbot-qdrant \
  -p 6333:6333 \
  -v "$(pwd)/.docbot/qdrant_storage:/qdrant/storage" \
  qdrant/qdrant
This command:
  • Runs the container in detached mode (-d)
  • Names it docbot-qdrant for easy reference
  • Exposes port 6333 for API access
  • Mounts a volume to persist data in .docbot/qdrant_storage
4

Verify Qdrant is running

Check the container status:
docker ps --filter name=docbot-qdrant
Test the health endpoint:
curl http://127.0.0.1:6333/health
You should see a response like:
{"status":"ok"}

​
Qdrant collections

Docbot creates two collections for indexing:

​
Documentation collection

Stores document chunks with the following schema:
  • Vector size: 1536 (default for OpenAI embeddings)
  • Distance metric: Cosine similarity
  • Payload indexes:
    • path (keyword) - Document file path
    • section (keyword) - Section heading

​
Code collection

Stores code chunks with additional metadata:
  • Vector size: 1536
  • Distance metric: Cosine similarity
  • Payload indexes:
    • path (keyword) - Source file path
    • symbol (keyword) - Function/class name
    • language (keyword) - Programming language
Collection names are automatically generated from your projectSlug configuration: docbot_{slug}_docs and docbot_{slug}_code.

​
Configuration

Configure Qdrant settings in docbot.config.jsonc: 
{
  "qdrant": {
    "url": "http://127.0.0.1:6333",
    "manifestPath": ".docbot/manifest.json",
    "collections": {
      "docs": "docbot_my-project_docs",
      "code": "docbot_my-project_code"
    }
  }
}

​
Configuration options

  • url - Qdrant instance URL (default: http://127.0.0.1:6333)
  • manifestPath - Path to index manifest file (tracks indexed files)
  • collections.docs - Name for documentation collection
  • collections.code - Name for code collection

​
Managing Qdrant

​
Starting and stopping

Stop the container: 
docker stop docbot-qdrant
Restart the container: 
docker start docbot-qdrant
Remove the container (data persists in volume): 
docker rm docbot-qdrant

​
Viewing logs

Check Qdrant logs for debugging: 
docker logs docbot-qdrant
Follow logs in real-time: 
docker logs -f docbot-qdrant

​
Accessing the Web UI

Qdrant includes a web interface at: 
http://127.0.0.1:6333/dashboard
Use it to:
  • Browse collections
  • Inspect vectors and payloads
  • Test search queries manually
  • Monitor collection statistics

​
Using a remote Qdrant instance

For CI/CD or multi-user scenarios, you may want to use a hosted Qdrant instance.
1

Deploy Qdrant

Options for hosted Qdrant:
  • Qdrant Cloud - Official managed service
  • Self-hosted on your infrastructure
  • Docker Compose for team development
2

Update configuration

Point Docbot to your remote instance:
{
  "qdrant": {
    "url": "https://your-qdrant-instance.com:6333"
  }
}
Or use the CLI flag:
docbot index --qdrant-url https://your-qdrant-instance.com:6333
3

Configure authentication (if needed)

If your Qdrant instance requires authentication, you’ll need to set up API keys according to your hosting provider’s documentation.

​
Data persistence

​
Local storage

The mounted volume at .docbot/qdrant_storage contains all your vector data. This directory:
  • Persists across container restarts
  • Should be added to .gitignore
  • Can be backed up for disaster recovery

​
Clearing and rebuilding

To completely reset your index: 
# Stop and remove container
docker stop docbot-qdrant
docker rm docbot-qdrant

# Remove storage
rm -rf .docbot/qdrant_storage

# Start fresh
docker run -d --name docbot-qdrant -p 6333:6333 \
  -v "$(pwd)/.docbot/qdrant_storage:/qdrant/storage" qdrant/qdrant

# Re-index
bunx @helmlabs/docbot index --force
Deleting qdrant_storage removes all indexed data. You’ll need to re-index your documentation and codebase.

​
Automatic initialization

When you run docbot init, Docbot automatically:
  1. Checks if Docker is installed
  2. Verifies if Qdrant is already running
  3. Attempts to start a container if needed
  4. Waits for Qdrant to become accessible
  5. Provides fallback instructions if automatic setup fails
You can skip this with: 
docbot init --skip-docker

​
Next steps

Build docs developers (and LLMs) love

Get started for freeTalk to us