Semantic and episodic memory as a service. Use one API endpoint and one key across computers, servers, IDEs, and CLI agents so every tool uses Jarvis as its system memory.
AI without memory starts from zero every time. Jarvis Memory gives your assistants shared context that survives tools, machines, and sessions.
Vector search over project context so agents can retrieve relevant facts from previous work in milliseconds.
Store events, decisions, task states, and conversations as timeline records that can be recalled later.
Cursor, Claude Code, OpenCode, Windsurf, Continue, Cline, Roo Code, LM Studio, Ollama, VS Code, and more.
The installer configures supported tools and writes a system memory bootstrap so agents search and save to Jarvis by default.
Project memory is isolated by API key. Keys can be copied to another machine without sharing passwords.
Download an open archive with README.md, manifest.json, memory.jsonl, and episodic.jsonl for migration to compatible services.
Register, copy the API endpoint and key, then run the setup command on every machine that should use Jarvis as primary memory.
One shared memory, three practical steps.
Register and receive an API key for your project memory.
Run jarvis-setup to configure MCP plus global memory rules for supported CLI and IDE agents.
Agents save and retrieve memory automatically, no matter which machine they run on.
Simple monthly access while the service is in early production.