The CheckFlow Copilot integration for GitHub Copilot and Copilot Studio.
Ask Copilot in VS Code what is due, raise a task for the pull request you are reviewing or complete a step in the release checklist, without leaving the editor. Copilot Studio agents can use the same tools.
MCP server on every planUses no AI creditsActs only as you
Prompted for the key once, stored securely
{
"inputs": [{
"type": "promptString",
"id": "checkflow-api-key",
"description": "CheckFlow API key",
"password": true
}],
"servers": {
"checkflow": {
"type": "http",
"url": "https://api.checkflow.io/mcp",
"headers": { "X-API-KEY": "${input:checkflow-api-key}" }
}
}
}Last updated
Copilot connects to CheckFlow through the CheckFlow MCP server, which signs in with an API key in a header. GitHub Copilot in VS Code, the Copilot CLI and the Copilot cloud agent can all send that header, and so can Copilot Studio agents. Copilot can then read your tasks, start checklists and complete steps as the member whose key it holds. Microsoft 365 Copilot plugins can’t connect yet.
Process work, from where you write code.
Stay in the editor
Check the release checklist, tick off a step or raise a follow-up task from Copilot Chat, without switching to the browser.
Agents your colleagues can use
A Copilot Studio agent can start and check on CheckFlow processes for people who never open VS Code.
Your permissions, not more
Copilot acts as the member whose key it holds. It sees what they see and is refused where they would be.
Today
Which Copilot can connect
“Copilot” covers several products. What matters is whether each one can send CheckFlow’s X-API-KEY header to a remote MCP server.
| Copilot | Connects? | How |
|---|---|---|
| GitHub Copilot in VS Code | Yes | A server entry in .vscode/mcp.json with a headers block; used in agent mode |
| GitHub Copilot CLI | Yes | copilot mcp add with the --header option |
| Copilot cloud agent | Yes, tools only | The repository’s MCP configuration, with the key in a COPILOT_MCP_ secret |
| Copilot Studio agents | Yes | The MCP wizard’s API key option, sent in a header you name X-API-KEY |
| Microsoft 365 Copilot plugins | No | MCP plugins for declarative agents don’t support API keys |
| Microsoft Copilot app | No documented way | Microsoft documents no custom MCP connections for it |
On Copilot Business and Copilot Enterprise, an organisation owner must switch on the “MCP servers in Copilot” policy first. It is off by default.
VS Code
Set up GitHub Copilot in VS Code
- 01
Get a key that acts as you
Ask an Administrator to create an API key on the API & Integrations tab of Team Management, acting as you, so that “my tasks” means yours.
- 02
Add the server
Create
.vscode/mcp.jsonin your workspace with the CheckFlow entry shown above. The inputs block makes VS Code ask for the key once and store it securely, rather than in the file. - 03
Start it
Start the server from the file, or run MCP: List Servers from the command palette, and enter the key when prompted.
- 04
Ask in agent mode
Open Copilot Chat in agent mode and ask “Which CheckFlow workspace are you connected to?”. It answers from
get_workspace_info.
Use https://api-eu.checkflow.io/mcp if your workspace is in the EU. A key sent to the wrong region is refused with a message naming the right address.
Terminal and cloud
The Copilot CLI and the cloud agent
In the Copilot CLI, add CheckFlow once with copilot mcp add and the --header option. It is saved to ~/.copilot/mcp-config.json and available in every session.
For the Copilot cloud agent, a repository admin adds CheckFlow to the repository’s MCP configuration, with the key stored as an Agents secret whose name starts with COPILOT_MCP_. GitHub recommends listing only the tools the agent needs, so start with read-only ones such as list_my_work and get_checklist.
- The cloud agent supports tools only, not resources
- Give the agent its own key, named for it, so you can revoke it alone
- Keep the key out of files you commit
Copilot CLI
copilot mcp add --transport http \
--header "X-API-KEY: your-api-key-here" \
checkflow https://api.checkflow.io/mcpCloud agent · repository MCP configuration
{
"mcpServers": {
"checkflow": {
"type": "http",
"url": "https://api.checkflow.io/mcp",
"headers": { "X-API-KEY": "$COPILOT_MCP_CHECKFLOW_KEY" },
"tools": ["list_my_work", "get_checklist"]
}
}
}Copilot Studio
Connect a Copilot Studio agent
- 01
Add an MCP tool
On your agent’s Tools page, choose Add a tool, then New tool, then Model Context Protocol.
- 02
Describe the server
Enter a name, a short description of what CheckFlow does (the agent uses it to decide when to call it) and the server URL,
https://api.checkflow.io/mcp. - 03
Choose API key in a header
Pick API key as the authentication type, Header as where it goes, and
X-API-KEYas the header name. - 04
Connect with a key
Create the connection with a CheckFlow API key. Microsoft notes that the agent’s user provides the key, so use a key that acts as that person.
Copilot Studio uses the Streamable HTTP transport, which is what the CheckFlow MCP server speaks.
Honestly
What Copilot can’t do in CheckFlow
- 01
Reach further than its key
Copilot is refused wherever its member would be: a step assigned only to someone else, a task behind an incomplete halt, a template they can’t edit.
- 02
Connect from Microsoft 365 Copilot
Its MCP plugins sign in with OAuth or Microsoft Entra, not an API key, and CheckFlow doesn’t support OAuth yet.
- 03
Report on analytics
Cycle times and bottlenecks come from Analytics in the app. The MCP server has no analytics tool.
- 04
Act as a separate user
Tasks are assigned to people and groups. Copilot works as the member whose key it holds, and its changes are recorded in their name.
The MCP server is included on every plan, starting with Business at $10 per user per month, or $9 billed annually, and in the free trial. Copilot runs on your own Copilot plan, so it uses no CheckFlow AI credits.
Ask, and it happens in CheckFlow.
An example request from VS Code, the real tool Copilot calls, and what it can truthfully say back.
Claude too
Claude Code and Claude Desktop connect to the same MCP server with the same API key.
Claude and CheckFlowDraft a template first
The free AI checklist generator drafts a release or deployment checklist from one sentence.
Try the AI generatorBuild it yourself
REST API v3 and signed webhooks give your own tools the same operations.
Developer overview
Raise a CheckFlow task for Priya to review the database migration PR by Friday, attached to this week’s release checklist.
name: "Review database migration PR" assignees: "Priya N." dueDateTime: "2026-10-16T17:00:00Z" attachedChecklistKey: "Software Rollout – 4.12"
Done. Priya has the review task, due Friday at 17:00 UTC, and has been notified. It is attached to Software Rollout 4.12, so that checklist won’t complete until the review is done.
Templates engineering teams run with Copilot.
Check on them and complete their steps from the editor.
Related pages
Copilot questions
Does CheckFlow work with GitHub Copilot?
Yes, through the CheckFlow MCP server. GitHub Copilot in VS Code connects with a server entry in .vscode/mcp.json and uses the tools in agent mode. The Copilot CLI and the Copilot cloud agent can connect too, because both can send the API key header.
Does CheckFlow work with Microsoft 365 Copilot?
Not directly yet. Microsoft 365 Copilot’s MCP plugins sign in with OAuth or Microsoft Entra rather than an API key, and the CheckFlow MCP server doesn’t support OAuth yet. A Copilot Studio agent can connect with an API key header instead.
Can a Copilot Studio agent use CheckFlow?
Yes. In the agent’s Tools, add a Model Context Protocol tool with the URL https://api.checkflow.io/mcp, choose API key authentication sent in a header named X-API-KEY, and create the connection with a CheckFlow API key.
Why can’t I see the CheckFlow tools in Copilot at work?
On Copilot Business and Copilot Enterprise, an organisation owner has to switch on the “MCP servers in Copilot” policy, which is off by default. If it is on, check the key: the header must be exactly X-API-KEY, and EU workspaces use api-eu.checkflow.io.
What can Copilot do in CheckFlow?
Whatever the member its key acts as can do, through 147 tools: read the Tasks Grid, start checklists, complete tasks and fill in fields, raise standalone tasks, build templates, update Data Sets and manage schedules and webhooks. It can’t report on analytics, and it is refused wherever its member would be.
Does using Copilot with CheckFlow use AI credits?
No. Copilot runs on your own Copilot plan. MCP calls use no CheckFlow AI credits; they count towards your workspace’s hourly API request budget, shared with the REST API.
Bring your checklists into the editor.
Start a free trial, create an API key that acts as you and connect GitHub Copilot in minutes.
14-day free trial, no credit cardFull refund in your first 30 daysUS or EU hosting