End-to-end memory infrastructure for AI.
Give agents, copilots, applications, and workflows persistent memory that survives sessions, context windows, and model changes.
## Why Memory
AI systems are powerful inside their current context window, but that context is temporary. Important decisions, user preferences, previous interactions, and learned knowledge disappear between sessions.
A larger prompt is not the same as memory.
Real memory must be:
- Persistent across sessions and model changes
- Selective about what deserves to be remembered
- Grounded in sources and provenance
- Queryable through semantic and structured retrieval
- Inspectable by the people operating the system
- Governed with clear access and retention controls
MemCode is building this layer as infrastructure rather than another temporary context workaround.
- Node.js
- npm
npm installCopy the provided environment template:
cp .env.example .env.localUpdate the values for your local or hosted environment. Do not commit production credentials or private keys.
npm run devnpm run buildRun the focused dashboard checks with:
npm run test:byok-usage
npm run test:runtime-settingsThe production build also runs TypeScript validation before generating the Vite bundle.
The dashboard is configured for deployment through Cloudflare:
npm run build
npx wrangler deployProduction environment variables should be configured through the deployment platform rather than committed to the repository.
The primary dashboard is available at app.memcode.in.
MemCode is under active development. APIs, interfaces, and infrastructure may evolve as we improve retrieval quality, memory evaluation, integrations, and developer experience.
We are building toward a world where every AI system can remember selectively, retrieve reliably, and improve continuously.