MCP and API keys

MCP is the way AI tools connect to other products. With Volt connected, you can ask your AI tool to find issues, create and update them, read and write docs, review pull requests and log time, without leaving it.

Connect an app

Open Settings › MCP and API keys and press Connect app. The page shows the server’s address and how to add it to each app:

• Claude: Settings → Connectors → Add custom connector, paste the address and sign in to Volt.
• ChatGPT: add a custom plugin with the address and sign in to Volt.
• Cursor: add a server with the address in Cursor’s MCP settings and sign in to Volt.
• Claude Code: run the command the page shows, then type /mcp in Claude Code and sign in to Volt.
• Other: any agent that supports MCP with OAuth takes the same address.

When you sign in, Volt asks which workspace the app works in and what it may do there: read only or read and edit, in all of your projects or only in some.

API keys

For an agent that can’t sign in, such as a script or a deploy bot, press Create key.

• Name. It shows on the agent’s changes, like “via Deploy bot”.
• Access. Only read, or read and edit.
• Projects. All of your projects, or the ones you choose.
• Expires. Never, or in a number of days.

Copy the key when it is made: Volt doesn’t show it again. Revoke takes a key’s access away at once.

What an app can do

The app acts as you: it sees what you see and can change what you can change, within the access you gave it. Its changes are marked with its name, like “Alex via Claude”, so the team knows who did what.

Two things are off unless you turn them on for a connection:

• Read meeting transcripts, of the meetings you can see.
• Manage projects and invite people: create and set up projects, add members, invite to Volt.

Deleting always needs your word: the request comes to your Inbox, where you confirm or decline it.

The app also follows the guidance written for Volt AI in the workspace and its projects, and can use your skills: see Volt AI.

Commands

Volt adds ready-made commands to the app. In Claude Code they start with /volt:, in VS Code with /mcp.volt; Claude and Cursor show them in their commands menu.

Command	What it does
work_on_issue	Pulls the issue, its discussion and links, plans, and starts working
my_day	Your issues in progress and up next, new mentions, a plan for today
standup	Done, in progress and blocked in a project, for a period
triage	Suggests priority, assignee and labels for new issues
break_down	Splits an issue into sub-issues once you agree
file_bug	Turns a description or a log into an issue, after checking for duplicates
plan_sprint	Picks the next sprint from the backlog
sprint_review	What got done, what slipped, what to carry over
release_notes	Drafts release notes from closed issues
write_spec	Writes a spec doc: problem, proposal, open questions
setup_project	Sets up a project with statuses, labels, members and invites
meeting_notes	Decisions and action items from a transcript

Disconnect an app

Connected apps lists the apps that have access, and each one’s activity: the calls it made in the last 90 days, without their content. Disconnecting an app takes its access away on every device; the changes it made stay.

For admins: agent access

Settings › Agent access decides what agents may do in the workspace as a whole.

• Allow agents over MCP. Off stops every app and key in the workspace.
• Meeting transcripts. Whether members may let their agents read transcripts.
• Projects and invites. Whether agents may set up projects and invite people, within the member’s role.

Below are the Connections of every member, each of which an admin can revoke, and the Activity of all of them for the last 90 days.