Think inside your AI world.
How do I get Claude to know my roadmap?
Four working answers, in ascending order of how well they survive the one thing a roadmap reliably does: change.
You can paste it, attach it to a Project, write it into CLAUDE.md, or connect it live. All four genuinely work on day one. They differ on day thirty, because a roadmap is not a document — it is a stack of decisions that keep getting revised, and each method has a different answer to 'which version is Claude reading?'
The file-shaped answers
Pasting works for a single conversation and dies with it. A Claude Project holds instructions and files across chats — put the roadmap doc there and every conversation in that Project starts briefed. In Claude Code, a CLAUDE.md in the repo does the same for coding sessions, loaded at start-up. Both are the right home for the stable skeleton: the quarters, the themes, the conventions.
Their shared cost is grooming. The file says what the roadmap was when someone last edited it. The interesting part of a roadmap — what got pulled forward, what got cut, and why — changes between edits, and a stale roadmap in a Project reads exactly as confidently as a current one.
What a roadmap is made of
Strip the formatting and a roadmap is settled calls: ship X before Y, cut Z this quarter, hold the migration until the audit lands — each with a reason, each revisable. When Claude works from the document alone it gets the conclusions without the reasoning, so the first time you push back it re-argues positions you already closed. Knowing the roadmap and inheriting its decisions are different depths of knowing.
The live answer
Unl holds the roadmap as those decisions: each call with its why, superseded cleanly when you change course. Connected to Claude or Claude Code over MCP, the current state is served into the session when it bears on the work — no re-paste, no stale file, and when a call was revised last Tuesday, what arrives is Tuesday's call, not March's document.
In practice the strong setup is both layers: skeleton in the Project or CLAUDE.md, live decisions in Unl. The file tells Claude the shape of the road; Unl tells it what you have settled about the journey since.
Claude knows your roadmap when it inherits the decisions the roadmap is made of — current, with their whys — not when it has read a document that was true at the last edit.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. The reach lane is live: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Read further
Questions people ask
Is a Claude Project enough for a small, stable roadmap?
Often, yes. If the roadmap changes quarterly and one person owns the doc, a Project with the roadmap attached is a clean answer. The gap opens when the roadmap moves faster than the doc — the Project keeps serving the old version with full confidence, and nothing flags the drift.
Should the roadmap live in CLAUDE.md for Claude Code?
The stable parts, yes — themes, sequencing rules, conventions. CLAUDE.md is loaded at session start and is version-controlled with the repo, which suits things that change at repo pace. Decisions that move weekly do better in a layer that tracks supersession, so the agent inherits the current call rather than the last commit's.
What does 'served when it bears on the work' mean?
Unl doesn't dump the whole roadmap into every session. When the work touches a decision you settled — you start speccing the feature you deferred, say — that decision arrives with its reasoning, unprompted. The rest stays out of the window until relevant.
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.