About CortexPrism

The open-source agentic harness that puts powerful AI capabilities in your hands.

What is CortexPrism?

CortexPrism is a single-process agentic harness written in TypeScript/Deno. It exposes a CLI, a REST API + WebSocket server, and a web UI. All state is persisted in SQLite databases using WAL mode.

At its core, CortexPrism is an agent loop that orchestrates LLM calls, tool execution, memory operations, and reflection. It supports 12+ LLM providers through a unified interface, with a cascading model router for cost optimization.

The system features a 5-tier memory architecture with hybrid FTS5 keyword + vector embedding retrieval, a defense-in-depth security model (Parallax) with encrypted credential vault, and sandboxed code execution with auto-fix loops.

Why Deno?

Deno provides a secure-by-default runtime with TypeScript support out of the box, modern JavaScript APIs, and efficient process management. It allows CortexPrism to run as a single binary with minimal dependencies.

Why Open Source?

We believe AI infrastructure should be transparent, auditable, and community-driven. The MIT license ensures CortexPrism remains free for everyone — individuals, startups, and enterprises alike.

Provider Freedom

No vendor lock-in. Use any LLM provider — or multiple at once. Switch between Anthropic, OpenAI, Google, local Ollama models, or any of 12+ providers with a single flag.

Extensible

Built-in tool system, plugin architecture supporting ESM/MCP/WASM, and a full REST API. Extend CortexPrism with custom tools, plugins, and integrations.

Project Principles

Privacy First

All data stored locally. No telemetry, no tracking. Your conversations, memory, and credentials remain on your machine.

Security by Design

Defense-in-depth with Parallax security model. Every tool call validated, every access logged, credentials encrypted.

Provider Agnostic

Use any LLM provider. Switch freely. The unified interface means your workflows are portable.

Local by Default

Run entirely on your machine. No cloud dependency for core functionality. Optional connectivity for LLM APIs.

Community Driven

MIT licensed. Contributions welcome. The roadmap is shaped by the community.

Built with love for the AI community