Turning Strategy Into Shared Operational Context
Dotwork MCP provides secure, live access to your ontology, objects, properties, relationships, and history, so AI is grounded in the operating model and the organization can update context seamlessly.
Sign in to connect to the
Dotwork MCP
{
"mcpServers": {
"dotwork": {
"url": "https://<your-dotwork-host>/mcp"
}
}
}Dotwork resolves identity and workspace scope after the client connects to the MCP endpoint.

A live model of how your organization operates
Give AI the same operating structure your teams use: value streams, products, teams, goals, work, metrics, workflows, relationships, and ownership, all in a connected operational graph.
Grounded intelligence from real operating context
Help agents generate plans, summaries, recommendations, and decisions using current strategy, portfolio activity, dependencies, discussions, and execution data, not isolated prompts or stale documentation.
Operational actions with built-in guardrails
Let agents safely prepare updates, create relationships, move workflows forward, coordinate handoffs, and draft changes inside governed operational boundaries with human approval where needed.
Context that always links back to the source
Every answer connects back to the relevant Dotwork artifact, workflow, inbox, or operational record so knowledge stays connected to the system of work, not trapped in chat threads.
What teams use it for
Powering skills for visibility, exploration, and action
Connect your preferred MCP-compatible client, complete Dotwork sign-in, and give your assistant the context it needs to find, explain, update, and route work.

Headless Strategic Context, Built for the Enterprise.
Dotwork MCP inherits your authentication, permissions, site policy, and approval model. Agents can be helpful without becoming another unmanaged integration surface.
Frequently asked questions
What is MCP?
Model Context Protocol is a standard way for AI clients to connect to external systems. Dotwork uses it to expose the ontology and operational graph in a structured, permissioned way.
Who is Dotwork MCP for?
It is for teams that want both people and agents to understand strategy, portfolios, operating structures, dependencies and work history, instead of only reading disconnected app data.
Can it make changes in Dotwork?
Yes, when the client, site policy and user approval allow it. Access is governed, so write-oriented actions can stay confirm-first.
Which clients can connect?
Any MCP-compatible client that can reach the Dotwork endpoint and complete the supported sign-in flow. Common examples include Cursor, Claude Code, Codex, ChatGPT and Cowork.
What can I do with the Dotwork MCP server?
Search the model, follow relationships and read history; generate plans, summaries and recommendations from current strategy and portfolio data; and, where allowed, prepare updates, create relationships and move workflows forward with human approval.
How does sign-in work?
It uses the normal Dotwork sign-in flow, including SAML SSO, OAuth or basic auth when enabled. Access is scoped to the signed-in user, the active site, workspace policy and your deployment.
Does it respect Dotwork permissions?
Yes. Every call carries the person’s own permissions, so an assistant can only see and change what that person can.
Can humans approve actions before they happen?
Yes. You can keep people in the loop for approvals before higher-risk reads, writes, workflow starts or messages take action in Dotwork.
Can we see what AI usage through MCP costs?
Yes. Token spend is shown in detail across agent actions and workflows, so teams understand AI usage inside Dotwork.
Do answers link back to Dotwork?
Yes. Every answer connects back to the relevant Dotwork record, workflow or inbox, so knowledge stays connected to the system of work instead of trapped in chat threads.