Nemocode

A coding agent that keeps what it learns about your project. Nemocode runs in your terminal, works on your files, and builds a memory of the project that survives the session it was learned in.

It remembers

Decisions, constraints, what a document demands — kept with a reason and a pointer back to the source, and offered back on the next mission.

You decide what it keeps

Rules, decisions, deletions and everything the curator or another agent proposes wait at a gate for your click. You approve, reject, or argue it out with the curator.

It stops for the things that matter

Plans before building, approval before shell commands, and an optional second agent that checks the work before a mission may finish.

What a session looks like

You open a project, write what you want, and watch it work. Nemocode reads before it changes, shows you the plan, and asks before anything it is not allowed to do on its own. When the mission ends, it writes down what is worth remembering — and asks you before that becomes part of the project memory.

Where to go from here

If you want toRead
Get it runningInstall & licence
Understand the first runYour first mission
Know every commandCommands
Understand the memoryProject memory
Control costModels & cost
Set the safety limitsPermissions & safety
Share the memory with colleaguesSharing with a team
See what the agent can doWhat the agent can do
Use the terminal commandsThe nemocode command
Automate work (hooks, schedules, connectors)Automation
Fix a problemTroubleshooting
Use the memory from Claude Code, Codex, opencode or Claude DesktopOther coding agents

What leaves your machine

Your code goes to the model provider you configured, and nowhere else. The project memory, the session history and the code index stay on your machine. If you share the memory with a team, approved facts also go to the shared repository you named — and only there. Nemocode contacts our licence server for two things only: to check your licence, and to fetch an update when you ask for one.