Runsight

Runsight is a YAML-first workflow engine designed for AI agents. It allows users to design agent workflows visually, with the canvas writing YAML directly to your filesystem. This enables version control, code review, and the ability to manage workflows like any other code in your repository.

The main features of Runsight include its Git-native approach, where workflows are stored as YAML files in your repository, per-run cost tracking to understand the expenses associated with each agent call, and runtime intervention capabilities that allow users to pause, inspect, fix, and resume workflows while they are running.

Runsight is built for engineering teams that run AI agents in production. It caters to teams that require the same level of rigor for their workflows as they do for their code, including version control and the ability to manage processes effectively.

Currently, there are no user-generated pros and cons available for Runsight. However, its features suggest that it offers significant advantages for teams needing a robust workflow management system for AI agents, such as Git-native workflows and runtime intervention. Potential cons may include the learning curve associated with adopting a new system and the need for self-hosting.

Yes, Runsight is open source and can be self-hosted, allowing users to customize and manage their workflow engine according to their specific needs.