I’ve been working on Runsight a YAML-first workflow engine for AI agents.

I’ve been working on Runsight a YAML-first workflow engine for AI agents. It lets you design agent workflows visually, while the canvas writes clean YAML directly to your filesystem. That means you can commit, diff, and review workflows in PRs just like any other code. And when something breaks at 2 AM, you don’t have to restart and hope for the best. Pause the running agent, fix the prompt, and resume no redeployment needed. What makes it different: → Git-native — workflows live as YAML in your repo, not hidden in a database → Per-run cost tracking — know exactly what each agent call costs → Runtime intervention — pause, inspect, fix, and continue while it’s running Built for engineering teams running agents in production who need real rigor: version control, code review, and the ability to stop a runaway process. Check https://runsight.ai/ Open source. Self-hosted.