Connect your AI agent

Let Claude Code, Codex, Cursor or VS Code see your games, their builds and what your testers said, and publish if you allow it.

Create an account

Building in the Playground? Its assistant is already connected; this page is for the agent in your own editor.

Make a token

Team settingsDeveloperAPI tokens, then Create token. Pick AI agent (read-only) to read games, builds, analytics and feedback, or Build pipeline if the agent should publish builds. Keep it in your shell as BOARDFLIGHT_TOKEN:

export BOARDFLIGHT_TOKEN=bft_…

Why one token cannot do both

Playtester feedback is written by people you handed a build to, so an agent reading it is reading a stranger's words. A token that reads feedback therefore cannot publish, promote or grant access (BoardFlight refuses to create one that could), and an agent connected with it is not offered those tools.

Feedback screenshots and redeem codes are never sent to an agent. A token is shown once; revoke it from the same list whenever you like.

Add the server to your agent

One command; drop --scope user to add it to this project only.

claude mcp add --transport http --scope user boardflight \
  https://mcp.boardflight.app/mcp \
  --header "Authorization: Bearer $BOARDFLIGHT_TOKEN"

Ask it something

Start a new session so the agent loads the server, then try:

“What did we publish to beta this month, and when?”
“Summarise what testers reported on the latest beta build.”
“How long do people play the newest build compared with the one before?”

What the agent can do

Either token reads your team's games, their channels and build history, and which glyph is which Piece. An AI agent (read-only) token adds play-session analytics and playtester feedback, marked as text a tester wrote. A Build pipeline token adds uploading and publishing a build, the same three steps as the CLI. Promoting builds and managing who has access stay in the dashboard.