Agentic Tools

Qdrant provides tools and resources that help AI agents work with Qdrant more effectively. Use them to give your agent task-specific guidance, troubleshoot existing deployments, design and evaluate search systems, and apply Qdrant best practices to your specific use case.

Start Building

If you’re new to Qdrant, the Start building on Qdrant prompt takes your coding assistant from an empty project to a working search query. It sets up a free Qdrant Cloud cluster (recommended) or a local Docker instance, installs the client and the Qdrant Advisor skill, and then either runs a quick demo on sample data or builds against your own use case.

Agent Prompt Start building on Qdrant

Help me get started building on Qdrant.

1. Ask whether I want a free Qdrant Cloud cluster or a local Docker instance. Recommend Cloud: nothing needs to run on my machine, and Cloud Inference creates embeddings for free.

2. Create a new, empty directory named quickstart-YYYY-MM-DD with today's date, and work in it. Install the Qdrant Advisor skill there: npx skills add qdrant/skills/meta/qdrant-advisor

3. Cloud: ask me to create a free cluster at https://cloud.qdrant.io. When it's created, the console shows the cluster's endpoint and an API key. Ask me to copy both into a .env file in the directory as QDRANT_URL and QDRANT_API_KEY, and load them from there. Never ask me to paste them in chat. Add .env to .gitignore.
   A new key can take a few seconds to work. If the first request returns 403, retry.
   Local: docker run -d -p 6333:6333 -p 6334:6334 -v "$(pwd)/qdrant_storage:/qdrant/storage:z" qdrant/qdrant

4. Ask which language I want, then install the Qdrant client.
   Cloud: no embedding library. Send text as Document objects with a free model from https://qdrant.tech/documentation/cloud/inference/#qdrant-hosted-models. In Python, set cloud_inference=True. For other languages, find examples with https://skills.qdrant.tech/snippets/search?language=<LANGUAGE>&query=cloud+inference
   Local: if I have no preference, use Python with pip install "qdrant-client[fastembed]".

5. First, offer a quick demo on sample data. If I decline, go to step 6. If I accept, download demo.py and menu-items.json from https://raw.githubusercontent.com/qdrant/examples/refs/heads/master/ai-getting-started/ and run python demo.py. If I picked another language, port demo.py to it, with the same data and queries.
   Cloud: install qdrant-client, then change only the client setup to use QDRANT_URL, QDRANT_API_KEY, and cloud_inference=True, and set EMBED_MODEL to sentence-transformers/all-minilm-l6-v2.
   Local: install qdrant-client[fastembed] and run it unmodified.
   When you show me the output, introduce it in a few sentences: what the dataset is, which text was embedded and with which model, what each query searched for, and why the top results match even when they share no words with the query.

6. Next, offer to build something on my own data. If I decline, stop. If I accept, ask what I'm building and what I'll search over before you write any code. If I don't have data yet, use menu-items.json, downloading it from the URL in step 5 if needed. Consult the Qdrant Advisor skill, and:
   - Take the vector size from the model. For Qdrant-hosted models, it's in the Dimensions column at https://qdrant.tech/documentation/cloud/inference/#qdrant-hosted-models. Never guess it.
   - Use the distance metric the model was trained for, and say why.
   - Create a payload index of the right type for every field I'll filter on, before loading data.
   - For multiple fields or hybrid search, explain the named-vector layout and its trade-offs.
   Then load a small sample, run a real query, and show me the results.

Uses the meta/qdrant-advisor agent skill. What are agent skills?

With Qdrant Cloud, nothing needs to run on your machine and you don’t need a credit card. Cloud Inference creates embeddings for free with the free Qdrant-hosted models, so you don’t need an embedding library or an embedding provider key.

Locally, Qdrant runs in Docker and FastEmbed generates embeddings on your machine, so you don’t need to sign up for anything.

Collection configuration is important. A mismatched distance metric or a missing payload index can result in slower or lower quality search. The prompt is written to make your assistant ask what you are building, then commit to those choices out loud before it writes any code.

Agent Skills

Agent skills encode solutions architect knowledge as structured files your agent loads on demand. They’re organized around symptoms, so they answer “why is this happening, and what should I check first?” Reach for skills when something is already running and behaving unexpectedly: slow search, growing memory, a stuck optimizer, poor relevance.

Prompts in the Docs

Short, ready-to-use prompts appear directly on the documentation pages where they are relevant, so you can use the prompts when you need them instead of looking for them elsewhere. Each prompt points to the agent skill your assistant should load. The prompt gets the task started; the skill provides the deeper Qdrant-specific guidance. The prompt index lists all available prompts.

Was this page useful?

On this page:

Ready to try Qdrant?