Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.veclabs.xyz/llms.txt

Use this file to discover all available pages before exploring further.

​
1. Get an API key

Sign up at app.veclabs.xyz. Your API key will be shown once — save it.

​
2. Install the SDK

npm install @veclabs/solvec

​
3. Create a collection

import { SolVec } from '@veclabs/solvec';

const sv = new SolVec({ apiKey: process.env.RECALL_API_KEY });
const collection = sv.collection('agent-memory', { dimensions: 1536 });

​
4. Upsert vectors

await collection.upsert([{
  id: 'mem_001',
  values: embedding,         // float[] of length 1536
  metadata: { text: 'User prefers dark mode' }
}]);
The response includes a merkleRoot — a SHA-256 fingerprint of your collection. On Pro and above, this root is posted to the Solana Anchor program automatically.

​
5. Query

const results = await collection.query({
  vector: queryEmbedding,
  topK: 5
});

// results.matches: [{ id, score, metadata }]

​
6. Verify (Pro and above)

TypeScript
const proof = await collection.verify();
console.log(proof.solanaExplorerUrl);
// → https://explorer.solana.com/tx/...?cluster=devnet
The verify call fetches the on-chain Merkle root from Solana and computes it locally from your collection. If they match, your memory is intact and tamper-evident.

​
What’s free vs paid?

FeatureFreePro ($25/mo)
Vectors5K500K
Writes/mo1K50K
Queries/mo10K500K
Irys permanent storage❌✅
Merkle root → Solana❌✅
Free tier vectors live in Redis cache. Pro tier vectors are stored permanently on Arweave via Irys and anchored to Solana on every write.

​
LangChain integration

npm install @veclabs/solvec @langchain/core

import { RecallVectorStore } from '@veclabs/solvec/langchain';
import { OpenAIEmbeddings } from '@langchain/openai';

const store = await RecallVectorStore.fromTexts(
  ["User prefers dark mode", "Meeting at 3pm"],
  [{ source: "chat" }, { source: "calendar" }],
  new OpenAIEmbeddings(),
  { apiKey: process.env.RECALL_API_KEY, collection: "langchain-memory" }
);

const docs = await store.similaritySearch("user preferences", 3);