What is the Empwr.ai MCP?
MCP (Model Context Protocol) is an open standard that lets AI assistants securely connect to other apps. The Empwr.ai MCP connects the assistant of your choice to your Empwr.ai workspace. You can ask about your projects, meetings, and tasks in plain language and get answers grounded in what actually happened. If you allow it, the assistant can also make changes for you, such as creating a task, assigning it, or marking it complete.
Before you begin
You'll need:
An Empwr.ai account that is a member of an organization. Accounts that are only guests in other organizations can't connect.
An account with the AI assistant you want to connect. Custom connector support varies by product and plan, so check your assistant's documentation if you aren't sure whether your plan supports custom MCP servers.
The Empwr.ai MCP URL
You'll be asked to paste this URL during setup in any of the assistants below:
Keep it handy. Every setup flow on this page uses the same URL.
What you're allowing
When you connect, Empwr.ai shows a consent screen that lists what the assistant will be able to do:
Access your meetings, projects, and tasks. Read access, always included.
Manage your projects and project-related work, including project setup, meeting associations, tasks, and risks. Write access, included when your assistant asks for it.
The assistant acts as you. It can only see and change what you can see and change in Empwr.ai, and its changes are recorded under your name. It can't delete projects, meetings, or tasks.
Some changes reach beyond the task or project itself. Changing a task's status is also sent to a linked Jira, Linear, or GitHub ticket. Linking a meeting to a project makes the meeting visible to the project's members and notifies them.
Step 1: Connect to your AI assistant
Pick the assistant you use and follow the steps below. After you approve access, the connection is live and ready for questions.
Claude (claude.ai, the Claude desktop app, and Claude mobile)
Connect Empwr.ai to Claude →
Empwr.ai is listed in Claude's Connectors Directory.
Sign in at claude.ai and open the Empwr.ai listing: https://claude.ai/directory/empwr-ai (or go to Customize → Connectors and search for "Empwr.ai").
Click Connect.
Sign in to Empwr.ai if asked, then click Allow on the Empwr.ai consent screen.
Once connected, Empwr.ai is available in your Claude conversations on the web, desktop, and mobile apps.
Empwr.ai is a Community connector, so Claude shows a short notice before you connect. That's expected for directory connectors that Anthropic hasn't verified.
On Claude Team and Enterprise plans, an Owner first adds Empwr.ai for your organization in Organization settings → Connectors. Each member then connects with their own Empwr.ai account. On Team plans, members who can't add it themselves will see a Request button that notifies their Owners.
Claude Code
If you're signed in to Claude Code with your claude.ai account, there's nothing extra to set up. Once you've connected Empwr.ai at claude.ai, it's available in Claude Code too. Run /mcp to see it, marked as coming from claude.ai.
If you use Claude Code with an API key or through a cloud provider, or just want to add it directly:
In your terminal, run:
claude mcp add --transport http empwr-ai https://app.empwr.ai/api/mcpIn Claude Code, run
/mcp, select empwr-ai, and choose Authenticate.Approve access on the Empwr.ai consent screen that opens in your browser.
ChatGPT
ChatGPT supports custom MCP connectors on eligible paid plans.
Sign in at chatgpt.com and open Settings → Connectors.
Click Create (or Add a connector, depending on your plan).
Paste https://app.empwr.ai/api/mcp as the MCP server URL.
Approve access when prompted.
OpenAI Codex
Codex supports custom MCP connectors on eligible paid plans.
Sign in to Codex and open Settings → MCP Servers.
Click Add Server.
Name the server Empwr.ai and paste https://app.empwr.ai/api/mcp as the MCP server URL.
Click Save and approve access when prompted.
If you don't see the Connectors option, your plan may not include custom MCP support. Email [email protected] and we'll point you to the right next step.
Microsoft Copilot
Microsoft Copilot supports MCP connectors through Copilot Studio agents.
Sign in at copilotstudio.microsoft.com.
Open or create the agent you'd like to add Empwr.ai to.
Go to Tools → Add a tool → Custom connector (look for the MCP server option).
Paste https://app.empwr.ai/api/mcp and complete the authorization.
If you're using the consumer version of Microsoft Copilot, custom connector support may be limited. Email [email protected] for guidance.
Step 2: Try a request
Once the connection is live, your assistant can pull context from your Empwr.ai workspace and, with write access, act on it. A few examples to get you started:
"What decisions and open questions came out of this week's meetings?"
"Give me the current status of the {project} rollout: blockers, risks, and open Jira issues."
"Create a task in {project} to send the pilot readiness checklist, assign it to me, and make it due Friday."
"Mark my task to update the launch plan as complete."
You can ask follow-up questions in the same conversation, and the assistant will pull fresh context from Empwr.ai each time. Many assistants ask you to confirm before they make a change.
Empwr.ai MCP tools
The Empwr.ai MCP server has 26 tools: 16 that read data and 10 that make changes. Your assistant picks the right tools for your request, so you don't need to name them.
Read tools
Tool | What it does |
list_projects | Lists the projects you belong to, with their goal, members, and latest status. |
get_project | Gets one project's details: goal, members, milestones, and latest status report. |
get_project_risks | Lists a project's risks and whether each is active or resolved. |
list_project_meetings | Lists the meetings linked to a project. |
list_project_creation_options | Shows which organization you can create projects in. |
get_project_documents | Lists a project's uploaded documents and linked Google Drive and SharePoint files. |
get_project_document_details | Reads an uploaded document. Linked files return a link instead of their content. |
list_project_message_transcripts | Lists the Slack and Microsoft Teams threads and forwarded emails linked to a project. |
get_project_message_transcript | Reads one of those threads or emails in full. |
list_project_integration_issues | Lists the Jira and Linear issues and GitHub issues and pull requests linked to a project. |
list_meetings | Lists meetings you have access to, from the past 28 days through the next 28 days unless you ask for another date range. |
get_meeting_outcome | Gets a meeting's AI-generated summary, decisions, action items, risks, and open questions. |
get_meeting_transcript | Gets a meeting's full transcript. If the meeting’s owner has hidden its outcome, only the owner can read it. |
list_tasks | Lists your tasks across projects, with status, assignees, and due dates. |
list_project_tasks | Lists the tasks in one project. |
list_action_items | Lists the action items from your meetings. |
Write tools
These tools need write access (see "What you're allowing" above).
Tool | What it does |
create_project | Creates a project in your organization, with you as its owner. |
update_project | Changes a project's name or goal. |
link_meeting_to_project | Links a meeting to a project. The project's members can then see the meeting and are notified, and Empwr.ai refreshes the project's status. |
unlink_meeting_from_project | Removes a meeting from a project. The meeting itself is kept. |
create_task | Creates a task, on its own or in a project you belong to. |
update_task | Changes a task's title, notes, or due date. |
assign_task | Assigns a task to one person, replacing any current assignee. |
update_task_state | Marks a task not started, in progress, or complete. The change is also sent to a linked Jira, Linear, or GitHub ticket. |
resolve_project_risk | Marks a project risk resolved, with an optional note. |
reopen_project_risk | Reopens a resolved risk and clears its resolution note. |
How they fit together
A typical request starts broad with list_projects, list_meetings, or list_tasks, narrows down with get_project, get_meeting_outcome, or list_project_tasks, and then acts with a tool such as create_task or update_task_state. The assistant usually chains these together on its own: you describe what you want, and it picks the tools.
Disconnecting and revoking access
To disconnect, remove Empwr.ai from your assistant's connector settings. In Claude, that's Settings → Connectors. In Claude Code, run claude mcp remove empwr-ai.
An assistant's sign-in to Empwr.ai expires 30 days after it was last used, so a connection you stop using ends on its own.
To change what an assistant can do, for example to add write access, disconnect it and connect again.
If you think an assistant has access it shouldn't, email [email protected] and we'll revoke it.
Troubleshooting
"You need to be a member of an organization to connect via MCP." Your account is only a guest in other organizations. Ask an admin of your organization to add you as a member.
Sign-in fails or keeps asking you to authorize. Remove the connector from your assistant, add it again, and approve access on the Empwr.ai consent screen. If that doesn't help, sign out of app.empwr.ai and try once more.
The assistant can read but can't make changes. The connection only has read access. Disconnect and connect again, and check that the consent screen lists "Manage your projects and project-related work". In Claude Code, also check that the connection doesn't use an old API key (see the Claude Code steps above).
A project, meeting, or document is missing. The assistant sees only what you can see in Empwr.ai. Check that you're a member of the project, or have access to the meeting, in Empwr.ai. For meetings more than 28 days old, ask for a specific date range. If a meeting’s owner has hidden its outcome, only the owner can see its summary and transcript.
"Too many requests." Empwr.ai limits how many requests and changes an assistant can make each minute. Wait a minute and try again.
A linked ticket didn't change. Status changes are sent to linked Jira, Linear, and GitHub tickets in the background, and a Jira workflow without a clear matching status can block the update. Other task edits, such as title, notes, due date, and assignee, stay in Empwr.ai.
Need help?
Custom MCP setup can look different depending on your AI assistant and your plan. If you hit a snag at any step, send us a note at [email protected] and we'll get you connected.