Cortex OSS
Open-source memory infrastructure for AI applications. Give your agents persistent, searchable, context-aware memory. No enterprise code, no license restrictions.
What is DoAide Cortex OSS?
Cortex OSS is a standalone copy of the open-source core from Cortex Platform. It contains everything you need to run a production memory system without any enterprise dependencies.
Write a memory in plain text. Cortex automatically extracts entities, builds a knowledge graph, detects conflicts with existing memories, and makes everything searchable via vector + full-text hybrid retrieval.
Deploy in 60 Seconds
git clone https://github.com/doaide/cortex.git
cd cortex
docker compose up -d
# Verify
curl http://localhost:3000/health
# → {"status":"ok"}That's it. PostgreSQL with pgvector, all Rust services, and NATS streaming — running locally.
Features
Hybrid Search
Vector similarity + BM25 full-text with Reciprocal Rank Fusion. Intent-aware routing picks the best strategy per query.
Knowledge Graph
Automatic entity and relationship extraction from conversations. N-hop traversal for rich contextual retrieval.
Multi-Scope Memory
User, session, agent, team, and org-level scoping. Row-level security isolation between tenants.
Memory Types
Episodic, semantic, procedural, and working memory. Four cognitive layers that mirror how humans remember.
PII Redaction
Built-in detection and redaction for emails, phones, SSNs, credit cards, and more. Privacy by default.
Conflict Detection
Automatic detection of contradictions and duplicates. Temporal validity windows resolve conflicting facts.
Decay Scoring
Older memories gracefully fade; recently accessed ones stay relevant. Configurable decay curves per memory type.
Policy Engine
Retention limits, access control, content filtering, and auto-redaction. Fine-grained control over memory lifecycle.
Webhooks
Subscribe to memory events: created, updated, conflicts detected. Build reactive workflows on top of memory.
Architecture
Gateway (:3000)
API routing, auth, rate limiting
Write Service
Episode ingestion & webhooks
NATS Stream
Async extraction pipeline
Graph Service
Entity extraction, conflict detection, embeddings
Retrieve Service
Vector + BM25 + graph fusion (RRF)
PostgreSQL + pgvector
Single database for all data
Integrate in Minutes
Python
from cortex import CortexClient
client = CortexClient(
base_url="http://localhost:3000",
api_key="your-api-key"
)
# Store a memory
client.add(
content="User prefers dark mode",
user_id="user-123"
)
# Hybrid retrieval
results = client.search(
query="user preferences",
user_id="user-123"
)TypeScript
import { CortexClient } from '@doaide/cortex-sdk';
const cortex = new CortexClient({
baseUrl: 'http://localhost:3000',
apiKey: 'your-api-key',
});
// Store a memory
await cortex.add({
content: 'User prefers TypeScript',
userId: 'user-123',
});
// Hybrid retrieval
const results = await cortex.search({
query: 'programming preferences',
userId: 'user-123',
});Cortex OSS vs Cortex Platform
| Cortex OSS | Cortex Platform | |
|---|---|---|
| License | Apache 2.0 | Apache 2.0 + BSL (enterprise) |
| Core services | All included | All included |
| Knowledge graph | ✓ | ✓ |
| Hybrid retrieval | ✓ | ✓ |
| Memory decay | ✓ | ✓ |
| SDKs (Python, TS, etc.) | ✓ | ✓ |
| Admin dashboard | — | ✓ |
| SSO / SAML / SCIM | — | ✓ |
| Audit logging | — | ✓ |
| Automated backups | — | ✓ |
| Best for | Self-hosters & contributors | Teams & enterprise |
Related Products
Start building with Cortex OSS
Three commands to a running memory system. Fully open source, no strings attached.