Integrations AutoGen

Multi-Agent

AutoGen Conversations. Verified Facts.

Teach/recall with cryptographic proof.

Get API Key →
12MCP tools
0.01mslookup
Freefor individuals
ed25519signed

The Problem

AutoGen agents talk. They assert facts in conversation. But nobody checks those facts against memory. When agents disagree, there's no arbiter. No receipt of what was said, when, and whether it was true.

Step 1
Connect
Step 2
Ingest + Extract
Step 3
Verify + Receipt

The Fix

npm install @verifiedstate/autogen

typescript
import { VerifiedAutoGenMemory } from '@verifiedstate/autogen'; const memory = new VerifiedAutoGenMemory({ apiKey: 'vs_live_...', namespaceId: 'autogen_research', }); // Agents teach and recall with receipts await memory.teach('The API rate limit is 1000 req/min'); const fact = await memory.recall('rate limit'); // fact.receipt.signature → "ed25519:..."

What You Get Back

✓ VERIFIED
receipt.json
{ "receipt_id": "rct_7f3a...", "assertion": "User prefers dark mode", "status": "verified", "confidence": 0.94, "signature": "ed25519:a8c3f1...", "verified_at": "2026-04-06T14:32:01Z", "chain_position": 42 }

Without VerifiedState vs. With VerifiedState

FeatureWithoutWith VerifiedState
Agent assertions Unchecked conversation Verified with receipts
Conflicting claims No resolution Auto-detected
Memory persistence Session only Permanent with provenance

FAQ

Which AutoGen version?

Works with AutoGen 0.2+ and the new autogen-agentchat package.

Related Integrations

Start building with verified memory.

Free: 50,000 assertions/month. No credit card. No trial expiration.