claudrunner

to move · Enter to open · Esc to close

Guides

Turn Trello cards into pull requests with an AI coding agent

Drag a card into your ready list, and a pull request comes back with the card moved to review. This guide connects claudrunner to a Trello board.

How it uses your board

Trello queues are lists. init reads your board and maps lists by meaning, not by name, because people rename lists:

Role Typical list
Ready “To do”, “Ready”, “Triage”
Review “In review”, “PR open”
Parked “Needs input”, “Blocked”
Filed by the bug sweep “Inbox”, “Intake”

On each run claudrunner fetches cards from the ready list, adds a running label to claim them, and re-reads each card to make sure no other run claimed it at the same moment — Trello has no atomic claim, so the re-read is what stops two runs from building the same card.

Then the AI coding agent implements the card, writes and runs tests, and a fresh reviewer attacks the diff before the pull request opens. The card moves to your review list with the pull request link in a comment. A card it cannot finish goes to your parked list with a note saying exactly why.

Set it up

/plugin marketplace add samiraklf/claudrunner
/plugin install claudrunner
/claudrunner:init

Choose Trello. init proposes the list mapping and stores the list ids.

Credentials, as CI secrets or environment variables — never in the config file:

Variable Where to get it
TRELLO_API_KEY trello.com/power-ups/admin, your Power-Up’s API key
TRELLO_TOKEN The token generated for that key

The shell orchestrator uses them; the AI agent never sees them.

Try a single card with /claudrunner:triage, read the pull request, then turn on the schedule init printed.

One board, several repositories

A Trello board often feeds more than one codebase. Give each repository its own label and its own claudrunner install. A card carrying two repository labels would be built twice, so claudrunner parks it with an explanation instead of guessing.

Labels carry meaning

Severity and kind are read from your labels, never guessed from the wording. The bug sweep files its findings with the same labels, so they sort the way the rest of your board does.

Next

Try it on your own repository

/plugin marketplace add samiraklf/claudrunner
/plugin install claudrunner
/claudrunner:init

Free and open source. It runs on your own Claude Code subscription, and nothing runs until you turn the schedule on. Getting started

Edit this page on GitHub