Tessl

What is Tessl?

Tessl is the package manager for agent skills. It helps developers evaluate skills and manage them as versioned, verified packages so issues like bad outputs are caught earlier in the workflow.

The page focuses on an evaluation step for skills, either by pointing Tessl to a public GitHub repository/skill definition or by running a local review from your machine. Tessl is designed for developers who want dependable skill packages in a shared registry.

Key Features

• Evaluate skills via a GitHub URL to automatically scan a repository and add the skill to the open registry.
• Publish evaluated skills to an open registry so other developers can find and install them.
• Run local skill reviews from the terminal to evaluate any local path, including private repositories and work-in-progress skills.
• Stays local during local reviews so results are only visible to you (useful for confidential or in-progress work).
• Versioned and verified skills and context to support professional workflows where correctness matters.

How to Use Tessl

1. Evaluate a public skill from GitHub: provide a repository URL. Tessl scans the repository and adds the evaluated skill to the open registry.
2. Review locally: run Tessl from your terminal using the command shown on the page:
   • npx tessl skill review ./<path to SKILL.md folder>
3. Follow the docs for evaluation details: the page notes that you should read the documentation to understand how the review/evaluation process works.

Use Cases

• Sharing a skill you found on GitHub: when you locate a public skill repository, you can evaluate it via its GitHub URL and have Tessl add it to the open registry for others.
• Curating a public repository for others to install: if you want a broader audience to discover and build on evaluated skills, Tessl provides a workflow that scans the repo and publishes the result to the registry.
• Validating a private skill before publishing: for skills in private repos, run a local review from your machine so you can check output quality without exposing results.
• Testing work-in-progress skills: if a skill is not ready to publish, local review supports evaluation of local paths while the work is still being developed.
• Reducing wasted compute from bad skills: use Tessl evaluation to catch problems earlier—such as skills that produce wrong output—so you avoid running them repeatedly and restarting work.

FAQ

• How do I evaluate a skill hosted on GitHub? Paste a public repository URL into Tessl; it will scan the repository automatically and add the skill to the open registry.

• Can I review skills that are not public yet? Yes. You can run a review locally from your terminal, including for private repositories and work-in-progress skills.

• What does “stays local” mean for local reviews? The page states that results are only visible to you when running the local review command.

• What command should I run to review locally? The page provides: npx tessl skill review ./<path to SKILL.md folder>.

• Where can I learn how reviews work? The page says to read the docs to learn how review evals work.

Alternatives

• General-purpose CI test pipelines: using unit/integration tests (or custom evaluation scripts) to validate skill behavior before sharing it; unlike Tessl, they may not automatically integrate with a shared skills registry.
• Direct repository-based sharing without a registry: sharing skills via GitHub alone can work, but it doesn’t provide an explicit evaluation-and-discovery workflow the way Tessl’s open registry does.
• Local-only evaluation tooling: running scripts or evaluators locally for skills can cover private and in-progress use, but may not include versioned, verified packaging and registry publishing.
• Other package managers for software components: traditional package managers manage code dependencies; Tessl is specifically positioned around versioned and verified agent skill packages and their evaluation.