← Back to Docs

Quickstart

Send your first request in under 60 seconds.

1

Get an API key

Sign up at /keys to get your sk-... API key. Free tier: 1,000 requests/day.

2

Extract your first answer

Send a URL and a question. Slaash returns only the relevant nodes — 99% fewer tokens than raw HTML.

cURL

curl -X POST https://api.slaash.ai/v1/extract \
  -H "Authorization: Bearer sk-your-key-here" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://en.wikipedia.org/wiki/Rust_(programming_language)",
    "goal": "when was Rust created and by whom"
  }'

Python

from slaash import Slaash

s = Slaash(api_key="sk-your-key-here")

result = s.extract(
    url="https://en.wikipedia.org/wiki/Rust_(programming_language)",
    goal="when was Rust created and by whom"
)

for node in result.nodes:
    print(f"[{node.role}] {node.label}")

JavaScript

import Slaash from 'slaash'

const s = new Slaash({ apiKey: 'sk-your-key-here' })

const result = await s.extract(
  'https://en.wikipedia.org/wiki/Rust_(programming_language)',
  { goal: 'when was Rust created and by whom' }
)

result.nodes.forEach(n => console.log(`[${n.role}] ${n.label}`))
3

Teach it (optional)

Tell Slaash which nodes had the right answer. Next time, they'll rank higher automatically.

curl -X POST https://api.slaash.ai/v1/learn \
  -H "Authorization: Bearer sk-your-key-here" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://en.wikipedia.org/wiki/Rust_(programming_language)",
    "goal": "when was Rust created and by whom",
    "node_ids": [142, 289]
  }'

Output formats

Every extract call supports 3 output formats:

JSON — structured nodes with role, label, relevance, causal_boost. Best for programmatic access.

Markdown — readable text with headings and links. Best for LLM consumption.

TOON — Token-Oriented Object Notation. 86% smaller than JSON. Best for minimal token budgets.

What's next?

REST API Reference — all 12 primitives with request/response specs.

Slaash Guide — how goal expansion, causal learning, and feedback work.

MCP Integration — connect to Claude Desktop, Cursor, VS Code.

Playground — try all primitives live in your browser.