Developers

Build on Waymarq

There is more than one way into Waymarq, and they suit genuinely different situations — an MCP server for the assistant you already talk to, a CLI for anything with a shell, a REST API for what you build yourself, signed webhooks for hearing about changes, and a git connection that needs no cooperation at all. Everything below actually exists and is what the app itself runs on — this isn't a roadmap page.

REST API

Create and update projects, phases, tasks, the CRM side, and the commercial side — everything the web app itself uses.

Webhooks

Get a signed HTTP request the moment a quote is accepted, a task finishes, or a client reports an issue — no polling.

MCP server

Connect Claude, Cursor or ChatGPT straight to your plan — nine tools, a URL and a token, nothing to install.

Bring your own AI agent

Point Claude Code, or anything else that can make HTTP requests, at the API and let it plan and build out a whole project.

Which one do you want?

They compose rather than compete — everything here goes through the same services, so a repository with a binding file, a git connection and an MCP server configured gets all three and they agree with each other.

Way in	Best for	What it needs	Docs
Just use itOpen the app and work.	Anyone. Start here.	An account.	In the app
Give it your specA brief in, a plan out.	Starting a project from a document rather than a blank board.	An AI provider key in your settings, or the free trial credit.	In the app
Let your commits do itThe tracker keeps itself current.	Teams who already write good commit messages and would rather not update two things.	A webhook in GitHub or Bitbucket.	In the app
Connect your AI assistantAsk it what to work on.	Anyone already working alongside an AI assistant.	A URL and a token. Nothing to install.	Read
Run a commandnpx @waymarq/cli	Automation, and agents that can only shell out.	Node 18.17+, a token, and a committed .waymarq.json.	In the app
Build against the APIThe whole surface, over HTTP.	Anything the other options do not cover, and anything you are building yourself.	A token.	Read

Quick reference

Some of this is written for a person to read, some for an agent to fetch and parse — they're labelled so you send the right one to the right reader.

For AI agents
/api-docs
The full REST API reference as raw markdown — fetch this first if you’re briefing an agent.
For humans
/api/documentation
The same reference as an interactive Swagger UI — browse endpoints and call them from the page.
For AI agents
/docs
The raw OpenAPI (JSON) specification, for generating a client or loading into your own tooling.
For AI agents
/mcp-docs
The MCP server reference as raw markdown — connecting, the nine tools, and how permissions apply.
For AI agents
npx @waymarq/cli
The command-line client on npm. No dependencies, --json on everything, works anywhere there is a shell.
For AI agents
/webhook-docs
The webhooks reference as raw markdown — events, payloads, and signature verification in three languages.
For AI agents
Organisation settings → API → Import guide
A prompt generator that builds a ready-to-paste brief for your own coding agent, once you’re signed in.

Getting started

Create a token under Organisation settings → API. Copy it — it's only shown once.
Fetch the reference at /api-docs if an agent is doing the work, or browse /api/documentation yourself.
Create a project with POST /api/v1/projects, then add phases and tasks underneath it.
See the REST API page for suggested end-to-end flows, not just individual endpoints.

Building something with it?

The API is young and shaped by the people using it. If an endpoint you need is missing, or something doesn't behave the way you'd expect, get in touch — that feedback genuinely changes what gets built next.

Get in touch