“I use GitHub projects because my internal IT wouldn't approve Linear, but it's the same scenario. I attempted to use github projects/issues as the actual task list at first through the MCP, and then eventually went back to in-repo task tracking, and just created a persistent gh projects subagent that occasionally mirrors/updates it with the repo's state.”
“I was doing basically this workflow and it worked pretty well, but meant that I had to have cmux + Linear MCP + an IDE opened at all times. I just ended up building my own harness in an OSS app called Volli to do this now.”
“I want to see the best dev automation setup:
- issues from helpscout tickets or Sentry issues
- creates issues in github/linear
- spawns Coding agent (codex, CC, Cursor, don't care)
- does some kind of QA
- Creates a PR
- Notifies in Slack”
“ai now writes almost half the issues in linear at some orgs. prs tripled. dev time still went up. more agent output doesn't mean faster shipping, it just moves the bottleneck to review.”
“It's a product of necessity: as most teams we started with Linear. But Linear is... well, linear. It works until the complexity of the codebase stops being linear.”