hn.today

Launch HN: StarSling (YC P25) – Self-improving CI for agent-volume repos

starsling.dev85 points40 comments
Screenshot of Launch HN: StarSling (YC P25) – Self-improving CI for agent-volume repos

StarSling is a GitHub Actions runner service aimed at teams whose rising code output - especially from coding agents - creates CI bottlenecks. Its central pitch is that faster hardware alone is not enough: AI agents should also keep improving the workflows themselves. Teams switch runners by changing a single `runs-on` label, leaving their existing jobs and commands intact. StarSling says its Ubuntu runners cost 33% less than comparable GitHub-hosted runners, with per-second billing, unlimited concurrency, and no charge for queue time. Its Adaptive Compute feature routes jobs to available sandboxes and can add CPU or memory while a job is running.

After a team installs the GitHub app, optimization agents inspect workflows, logs, and machine telemetry, then propose pull requests for changes such as better caching, parallelization, test sharding, and faster dependency installation. The site reports examples including Mastra’s test suite dropping from 29 minutes 56 seconds to 5 minutes 6 seconds, Better Auth’s end-to-end tests falling from 2 minutes 22 seconds to 1 minute 4 seconds, and Partcl’s queue time shrinking from 9.5 minutes to 35 seconds. StarSling also offers GPU runners and configurable AI pull-request reviewers. New accounts receive 2,000 free minutes; standard four-vCPU Linux runners are priced at $0.008 per minute. The proposition is a CI system that both executes jobs faster and uses agent-generated fixes to reduce future build time and cost.

Read on starsling.dev40 comments on Hacker News

Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.

More in Startups

The daily digest

Today's best Hacker News stories, summarized and screenshotted, one email a day.