Every single capability is also an MCP tool. Products, plans, decisions and time in one operational picture - and your agents work the same surface you do, while you stay the one who decides.
Free forever for one maker - 3 products, no card.
Gate LAUNCH-1 has 4 of 5 conditions met. The open one is the security review, unassigned for 6 days. Recommend holding - but this is your call.
288 tools, the same services the panel on the left is built from. An agent can read everything and propose anything - and is refused exactly where a person has to sign.
One backend. Two front doors. Nothing an agent does is invisible to you, and nothing you do is hidden from it.
From the first idea to the launch gate - and the same surface an agent works from.
One home that tells you what needs a decision, which products went quiet, and what changed - every row explaining itself in words and leading straight to where you act.
Time is its own layer. Initiatives, milestones, releases and dated tasks share one timeline, with the plan drawn against the actual that really happened.
The container between a product and its tasks: the large change you are betting will move it, tying an intended outcome to the epics and the gate that carry it.
Ship on real checkpoints. A release publishes only when its launch gate passes - never a flag someone flips - and the moment it shipped is recorded as a fact.
Drop a thought onto the timeline in five seconds; turn that same block into a task, epic, milestone or initiative later, its origin preserved.
A human-in-the-loop queue where agents advise and a person decides. A visual review carries typed mockups, anchored comments and advisory votes.
Long-form memory that survives a session hand-off, plus append-only named logs read from the tail - ask for the last three entries, not the whole file.
Real checkpoints with manual, automated and AI-assessed conditions. A refused transition names the gate, its unmet conditions and the exact next call.
Six checklists say what "done" means for deploy, auth, payments, security, tests and the store. makermap knows what must be checked, remembers who signed off and when, and says when that went stale. It never looked at your systems itself, and the record says so on every line.
Scoped tokens, one-call product context, cursor pagination, permission-aware discovery, and the project docs themselves searchable section by section.
Nothing in makermap carries its own private dates. Initiatives, milestones, releases and tasks all hang off one canonical timeline, which is why the plan and the actual can be drawn on the same row - and why an overdue item can turn critical and say so, instead of just sitting there.
An agent that is told "forbidden" learns nothing and retries forever. So every refusal in makermap names what stopped it, why, and the exact next call that would succeed - the same answer over REST and over MCP, because both are the same service.
Add as many agents as you like - they cost nothing to connect. A shared pool of credits covers the one thing that has a real machine cost.
For one maker, forever
For a maker with a portfolio
For a team that ships together
Semantic search runs an embedding model, so it costs 1 credit. Saving a document, reading context, moving a card, posting an advisory, appending to a log - free, and always will be. Your balance is a running sum you can read at any time, with every charge naming what it paid for.
Sign up, mint a scoped token, paste one block of config. Your agent has the whole product in front of it.
One field, one email when a place opens. No newsletter, and nothing else is asked.
{ "mcpServers": { "makermap": { "url": "https://api.makermap.dev/api/mcp", "headers": { "Authorization": "Bearer mm_pat_..." } } }}Works with any MCP client. The token decides what the agent may touch - discovery only lists the tools that token can call.