Quick Configuration

Choose the setup that matches your environment.

Before You Start

Confirm you have the account, endpoint, or API key required for Qdrant.
Start with minimum scopes and read-only access where possible.
Keep secrets in environment variables instead of hardcoding them in JSON.

Last verified: 2026-06-13

Environment: uvx with mcp-server-qdrant over stdio (Qdrant vector DB)

Install command: uvx mcp-server-qdrant

Local config

JSON

Local config json

{
  "qdrant": {
    "command": "uvx",
    "args": ["mcp-server-qdrant"],
    "env": {
      "QDRANT_URL": "${QDRANT_URL}",
      "QDRANT_API_KEY": "${QDRANT_API_KEY}",
      "COLLECTION_NAME": "my-collection",
      "EMBEDDING_MODEL": "sentence-transformers/all-MiniLM-L6-v2"
    }
  }
}

After copying

Check the Qdrant JSON before running it in OpenClaw, then verify the server with a real prompt.

Check config Submit config review Run verification prompt Browse related MCPs

Verify prompt

Store a short note about our deployment process, then retrieve it by semantic search.

Expected success output

The note is stored and returned by a semantic query against the collection.

Common Pitfalls & Fixes

⚠️ Watch out: API key authentication, collection setup, and vector dimension matching.
🔑 Always store API keys in environment variables, never hardcode them in JSON.
🛡️ Start with read-only scopes if available to verify connection safely.

Security checklist

Keep QDRANT_API_KEY in env; do not expose your Qdrant endpoint publicly without auth.
Scope COLLECTION_NAME to a dedicated collection for this assistant.

Common errors

Error	Cause	Fix
`Connection refused / URL error`	QDRANT_URL is wrong or unreachable.	Point QDRANT_URL at your instance (e.g. http://localhost:6333 or the cloud URL).
`403 Unauthorized`	QDRANT_API_KEY is missing for a cloud instance.	Set the Qdrant Cloud API key as QDRANT_API_KEY.
`Dimension mismatch`	EMBEDDING_MODEL differs from the collection's vector size.	Match EMBEDDING_MODEL to the collection, or create a fresh collection.

Example Prompts

Once connected, try these prompts to test capabilities:

Search for similar vectors in a collection using a text query.
List all collections and their point counts.
Add new points to an existing collection.

Verification Checklist

Run 1-2 real prompts to confirm Qdrant returns usable data.
Check that error messages are clear enough for troubleshooting.
Document the required scopes, dependencies, and env vars for future reuse.

Qdrant