AI coding has made CI a bottleneck, so we reworked ours to keep up

Linear redesigned its CI system to address the bottleneck created by faster AI-assisted code production, focusing on reducing pull-request wait times and runner costs even as its test suite nearly quadrupled. By moving from GitHub Actions to faster third-party runners, adopting the native TypeScript compiler tsgo, simplifying lint rules to avoid type-checking, and using Oxlint, it substantially accelerated core checks. The team also shortened critical-path jobs through minimal and resilient Git checkouts, removed unnecessary merge gates, reduced repeated setup by preinstalling dependencies, installing only relevant monorepo packages, avoiding inefficient caches, reusing database schema snapshots, and consolidating small checks. Test execution improved through better file balancing across eight Vitest shards and an opt-in shared-module configuration with safeguards for isolation-sensitive tests. These changes lowered PR wait time from over six minutes to just above five minutes, roughly halved runner time per test, reduced API shard execution time, and helped prevent CI from growing into an increasingly costly development constraint as Linear adds about 2,000 tests each week.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.