A question people ask
Managing context in a long Claude Code project
The longer the project runs, the more there is to hold — and the more each new session starts from scratch. The context that matters on a long project isn't the transcript. It's the decisions that have piled up, each with its reason.
On a long project, managing context by hand doesn’t scale. Unl holds the accumulating decisions — each with its reasoning, superseded as you change your mind — and serves the relevant one to Claude Code unprompted when it bears. The project compounds instead of resetting. You stay in command; free at launch.
Long projects are decision-heavy, not transcript-heavy
It isn’t a memory problem to solve — it’s a missing inheritance.
Weeks in, the real state of a project isn’t everything that was said — it’s the settled shape: the architecture you chose, the constraints you set, the paths you closed, where things stand. That set grows, and it’s exactly what a session window can’t carry.
Managing that isn’t a window-size problem. It’s a matter of holding the decisions somewhere durable and serving the relevant one when it bears — so the length of the project becomes an asset that compounds, not a weight that resets.
Compounding, not resetting
Each session inherits the settled shape and adds to it.
Because decisions are held with their reasoning and superseded cleanly, a long project accumulates a coherent record of what governs it. Every session starts on that and moves it forward — the opposite of starting cold.
How it works
Connect once. From then on, what you settled arrives on its own.
1 · Connect. Add the MCP connector in a couple of steps, about five minutes. It plugs into Claude, Claude Code, ChatGPT & Cursor as an MCP connector.
2 · Settle a decision. When you make a call, it’s held with its reasoning — on your word, never scraped. Change your mind and it’s superseded, the old one retired.
3 · It arrives unprompted. Next session, the decision that bears on the work is served into the model’s context the moment it’s relevant — you don’t fetch it, and you don’t re-explain it. The model inherits what you decided and builds from it.
Questions people ask
How do I manage context in a long Claude Code project?
Carry the decisions, not the transcript. On a long project the governing context is the accumulating set of settled decisions — architecture, constraints, dead-ends — each with its reasoning. Unl holds that set and serves the relevant decision unprompted when it bears, so each session inherits the settled shape rather than rebuilding it.
Isn't this a job for the context window or a rules file?
The window can't hold weeks of a project, and a hand-maintained rules file goes stale as decisions change. Holding decisions durably with their reasoning — and serving the relevant one when it's needed — scales with the project instead of against it.
Does the same set of decisions work across tools?
Yes — over MCP the accumulated decisions carry across Claude, Claude Code, ChatGPT and Cursor, so a long project stays coherent wherever you pick it up.
What this is
Think inside your AI world — you stay in command
Unlimitless (Unl to friends) holds what you've settled, reads what your tools are showing, and catches what's changed out in the world — and hands your AI whatever bears on the work, the moment it's needed, without you asking. The right thing, in front of the model, unprompted, with you in command of the call. So you keep moving toward what you set out to build, on top of everything you've already decided.
It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector. Quick to connect, in a couple of steps.
Unlimitless is open now to invited Alpha. Apply for the Beta waitlist to come in ahead of the full launch:
Alpha is invite-only · free at launch.