Extract what matters. Leave 90% behind.

Paste a large document — CLAUDE.md, API docs, a README, a knowledge base — and ask a question. Context Extractor returns only the relevant chunks within your token budget. No LLM required. Runs in milliseconds.

① Paste document② Type query③ Set token budget④ Get relevant excerpt

Sign in to use Context Extractor — free for all users.

Sign in →

Paste a large document and a query to extract only the relevant sections.

Common use cases

CLAUDE.md / .cursorrules

90% reduction

Your context file grows to 8k tokens. Before each task, extract only the relevant rules. Inject 800 tokens instead of 8,000.

API documentation

95% reduction

Large API docs with hundreds of endpoints. Query "how do I paginate" and get just the pagination section.

RAG without infra

No infrastructure

No vector database, no embeddings API. TF-IDF based — paste the document, query it, get the relevant chunks. Zero setup.

Use via API

curl https://mdmin.dev/api/v1/extract \
  -H "Authorization: Bearer mdmin_sk_..." \
  -H "Content-Type: application/json" \
  -d '{"document": "...", "query": "how does auth work", "max_tokens": 2000}'

Get your API key in Settings. Available on Pro.