Linear issues filed and triaged by an AI agent

Most Linear automation is about getting work into Linear with the right fields, and keeping it honest afterwards. MadMax does both inside one run: an AI agent reads a bug report, a support ticket or an alert, decides the team, priority and labels, and files the issue; later steps update its state or add a comment when something changes elsewhere. The run can also stop and show a person the draft before anything lands in your backlog. Describe it in plain English and MadMax builds the graph, wires your Linear connection and tests each step on real data before you activate.

What teams build

• A webhook from the error tracker runs an AI agent that writes a one-line title and a reproducible description, then create_issue files it on the platform team at the priority the agent chose and emails the reporter the ENG- link.
• A :ticket: reaction in Slack reads the thread, an agent drafts the issue, and an approval step shows the draft in Slack; on approve the issue is created and its identifier is posted back in the thread.
• Every Monday a schedule runs list_issues filtered to In Progress, a loop checks each one's last update, and comment_issue asks for a status on the stale ones before one digest goes to the team channel.

How it connects

Paste a personal API key from Linear (Settings, API, Personal API keys). MadMax sends it the way Linear's GraphQL API expects, raw in the Authorization header with no Bearer prefix, stores it encrypted and never shows it again. Rotating the key is a reconnect on the same connection, and no step needs rewiring. Each Linear step picks its connection in the Inspector; see managing connections.

Related

• AI agent step
• Approval step
• Managing connections
• Slack
• MadMax vs n8n