Developer Tools

Developer API & SDKs

Give any AI persistent memory in 10 minutes.

Full REST API with authentication and rate limiting. Official SDKs for JavaScript and Python. MCP server for Claude and Cursor. Give your agents, bots, and workflows a second brain they can actually use.

6 ways to integrate

Whether you're building in JavaScript, Python, or connecting through an AI framework — there's a native integration ready to go.

JavaScript / Node

npm install @neuralsnap/sdk

Python

pip install neuralsnap

MCP Server

Claude, Cursor, and any MCP client

REST API

Any platform, any language

CrewAI

Python multi-agent framework

LlamaIndex

RAG and agent memory

5 lines to persistent memory

Store knowledge, search semantically, and crystallize raw text — all with a clean, consistent API.

import NeuralSnap from '@neuralsnap/sdk';

const ns = new NeuralSnap({ apiKey: 'ns_...' });

// Store knowledge
await ns.snapshots.create({
  name: "SOPs Beat Prompts",
  type: "Conviction",
  core: "Written procedures produce 10x better agent output",
  confidence: 0.94,
});

// Search by meaning
const results = await ns.search("how to improve agent quality");

// Crystallize raw text into structured snapshots
const snapshots = await ns.crystallize({
  text: "In our planning meeting we decided to...",
  source_type: "meeting",
});

Simple API key authentication

Generate API keys from your dashboard. Each key scopes to your account and can be revoked instantly. Use Bearer token authentication on every request.

  • API keys generated in Settings → API Keys
  • Bearer token authentication on all endpoints
  • Keys scope to your account — one key accesses all your brains
  • Revoke keys instantly from the dashboard
  • Session tokens also supported for frontend apps

Rate limiting built for production

Fair usage rate limits with clear headers. Every response includes X-RateLimit-Remaining so you can throttle proactively. Higher limits on paid plans.

  • Free: 60 req/min reads, 10 req/min crystallization
  • Pro: 300 req/min reads, 60 req/min crystallization
  • Business: 500+ req/min, custom limits available
  • X-RateLimit headers on every response
  • Retry-After header on 429 responses

MCP Server for Claude & Cursor

The NeuralSnap MCP server gives Claude, Cursor, and any MCP-compatible client direct access to your brain. Search knowledge, store memories, and crystallize text — all from within your AI tools.

// In your Claude/Cursor MCP config:
{
  "mcpServers": {
    "neuralsnap": {
      "command": "npx",
      "args": ["-y", "@neuralsnap/mcp-server"],
      "env": {
        "NEURALSNAP_API_KEY": "ns_..."
      }
    }
  }
}

// Then in Claude:
// "Search my brain for what I know about scaling teams"
// → Returns relevant snapshots and memories

Use Cases

AI Agent Memory

One API call to store what matters. One call to recall it. Your agents stop being goldfish and start building real knowledge.

RAG Enhancement

Use NeuralSnap as a structured knowledge layer alongside your vector store. Better context = better answers.

Workflow Automation

Crystallize meeting transcripts, process documents, extract knowledge — all programmatically via REST.

Related Features

Neural Snapshots

Core data model for the API

Specific Knowledge

Knowledge ingestion endpoints

Brain Graph

Edges API for graph operations

Start building

Get your API key in 30 seconds and start integrating NeuralSnap into your stack.