upskill CLI
A Hugging Face tool: a strong model creates a skill from agent work traces, a weak one uses it and the evaluation shows the gain
- 762 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Runs models and jobs at a third-party service and spends keys and balance
- Trace data goes to model providers
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/upskillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs models and jobs at a third-party service and spends keys and balance
- Trace data goes to model providers
This is third-party code. Review the repository files before installing.
What it does
upskill generates and evaluates skills from agent work traces. An expensive teacher model creates a skill, a cheap one uses it, and the evaluation shows whether it helps. In version 2 the default configuration runs evaluations on Hugging Face Jobs. Install with `uv pip install upskill` or run through `uvx upskill`. Model keys are needed in environment variables (`ANTHROPIC_API_KEY`, `OPENAI_API_KEY` and others), and there is an MCP mode by a `uvx` config. The Hugging Face organization, Apache-2.0 license.
Who it is for. For people who write skills for agents and want to check whether a skill helps a weak model instead of judging by feel.
Good fit when
- You need to create a skill from agent work traces and check its benefit
- You need to compare a cheap model's result with and without the skill
- You need a skill evaluation tool from Hugging Face
Not a fit when
- You need a ready skill collection: upskill creates and evaluates but holds no set
- You cannot run jobs and models at a third-party service
Example request
Take the trace of the latest session, create a skill from it with a strong model and show how much it improved the weak model's resultLimitations
The tool runs models and, in the default configuration, jobs on Hugging Face Jobs, so it spends keys and balance and sends data to a third-party service. Model keys are needed. The result depends on the chosen teacher and student models.
How to disable. Remove the package with uv pip uninstall upskill and revoke the model keys you set.
FAQ
What does upskill do?
It generates a skill from agent traces with a strong model and evaluates whether it helps a weak one.
How do I install it?
uv pip install upskill or uvx upskill.
Where do evaluations run?
By the default configuration of version 2 on Hugging Face Jobs.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI that detects your project's stack and installs matching agent skills from a vetted registry in one command
A skill that designs, tests, evaluates and packages other agent skills, starting from architecture
A tool that builds tested skills from a real workflow and governs their lifecycle through your own marketplace
SkillCorpus
skillcorpus
The open part of the SkillCorpus pipeline: collects SKILL.md files from public repos, checks security and licenses, scores quality and matches skills to a task
A CLI installer and package manager for coding agent skills: one command installs skills into many runtimes and helps build your own
A set of CLI tools and a skill that give a text-only agent sight: image analysis, long-screenshot OCR and UI restoration
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| upskill (this entry) | yes | 762 | |
| Agent Skill Manager | no | 955 | a skill manager that scans before installing |
| MCPM | no | 1,012 | a package manager and registry for MCP servers |
| Library Skills | no | 812 | skills shipped by the libraries themselves |
README in short
The README describes upskill as a tool for generating and evaluating skills from agent traces, installed with uv, with model keys and an MCP mode. Apache-2.0 license.