Tessl is a development platform and package manager specifically designed for agent skills, which are described as the new unit of software for developers. The platform helps developers evaluate and optimize agent skills to improve AI agent performance and reliability.

Tessl enables developers to evaluate agent skills to determine if they help, provide minimal uplift, or even degrade functionality. The platform allows optimization of skills to help agents use products properly, build in codebases, and enforce policies. It provides quality signals for agent skills that are typically treated like simple Markdown files without versioning or updates.

The platform uses AI evaluations to assess agent skill performance. Through this evaluation process, Tessl was able to help optimize ElevenLabs' skills, resulting in doubling their agent success rate in using their APIs.

Tessl helps developers stop debugging agent output and start shipping quality code faster by providing evaluation capabilities. It addresses the problem of developers spending time course-correcting agents instead of shipping new features.

The platform targets developers building personal projects, maintaining open-source libraries, or developing with AI at work. It's particularly relevant for teams working with agentic workflows and those considering internal documentation as potential agent skills.