Building the Bucket Linear Agent

I’m intrigued by the potential of Linear agents to help with product development, especially the grunt work.

Asking around, some of the agents being worked on by folks I know include agents that:

- auto-update issue statuses based on GitHub activity.

- post daily summaries of Linear changes to Slack.

- discusses product requirements with you and creates Linear issues

- listens to Sentry error stream and auto creates linear issues for new crash signatures

- interprets GitHub events such as pull requests and deployment reviews, and propagates that context back to related Linear tickets

OK, I think I've deciphered this.

Bucket is a feature-flagging SaaS

Linear is an issue tracker SaaS

Linear Agents (also known as "app users" according to their developer docs https://linear.app/developers/agents) is a feature of Linear where you can define "agents" which can be added to issues and will then go and do something with those issues. It looks to me like Linear Agents don't even necessarily involve an LLM - they're webhook based, so you can @-tag an "agent" by name which will then trigger a webhook to a third-party somewhere (such as Bucket) who then go ahead and run custom code and interact further with Linear via their API.

(I collect definitions of "AI agents", and this is the first one I've seen in recent years that doesn't necessarily involve any LLM integration at all.)