buildbetter-app

SkillRank: Find and Evaluate AI-Agent Skills

SkillRank is a Rust CLI and public registry for finding, installing, and evaluating skills used by Claude Code, Codex, Copilot, and other agents. Its paired evals measure skill effects on token use, speed, and success, and it integrates with BuildBetter ZeroShot.

View on GitHub Compare
Stars
0
Forks
0
Stars, last 30 days
+1
Last commit
2 days ago
Contributors
4

Our take

Newless than six months old and changing fast

Health score

55/100

How it is scored
Activity
60
Community
36
Issues
53
Pull requests
71

+1 stars in the last 30 days

SkillRank is a promising, actively changing tool for teams that want evidence about agent skills, but its young project status and undetected test coverage call for a trial before relying on it broadly.

Good fit if

  • You use Claude Code, Codex, Copilot, or another agent that can work with SKILL.md packages and want to manage skills from a CLI.
  • You can run paired evaluations on your own agent and value effort metrics, not just pass rates.
  • You want the MIT-licensed CLI and core library, or need a local registry for development.

Look elsewhere if

  • You need a mature project with a long operational track record; SkillRank is new and changing quickly.
  • You need evaluation results to be official without meeting the project's reproduction and trust-tier criteria.
  • You intend to offer a competing hosted registry based on its registry or website code, which is under a separate license.
All health signals
Last commit2026-10-19 (2 days ago)
Commits, last 90 days58
Releases, last 12 months6 (latest v0.1.5, 2026-07-30)
Contributors4 (top contributor: 70% of commits)
Issues closed, last 90 days2
Pull requests merged, last 90 days39 (typically merged in 0 days)
Project age4 months

Checked on 2026-10-22 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

1

Forks

0

Change since Nov 21, 2026: 0 stars·0 forks

Overview

SkillRank is a Rust CLI and public registry for SKILL.md packages used by Claude Code, Codex, Copilot, and other AI agents. It addresses a gap in skill directories that rank by popularity: SkillRank can run paired trials on your own agent to show how a skill affects token use, speed, and task success. It also integrates with BuildBetter ZeroShot, which can recommend skills from coding sessions and track realized savings.

The core search, install, and local evaluation workflows do not require an account. Teams can use the CLI directly, expose its tools to Claude Code and Codex through MCP, or run a local registry. Publishing skills, ratings, reviews, and eval results requires an account.

Key Features

  • Search, inspect, install, and manage agent skills, with content-hash verification and a lockfile entry for installs.
  • Recommend skills based on detected repository stack.
  • Run paired skill-versus-no-skill trials on a pinned fixture repository, with verifier isolation.
  • Report task-level success, token, turn, time, and cost changes, while distinguishing missing cost data from zero cost.
  • Show published evaluation results under separate official, community-reported, and self-reported trust tiers.
  • Register SkillRank tools for Claude Code and Codex through an MCP server.
  • Serve a local registry with a seed catalog or a supplied catalog file.

Use Cases

  • An agent developer wants to compare whether a skill reduces time or cost, even if it does not change task success.
  • A developer wants to discover and install a framework-specific skill in a repository without creating an account.
  • A team using Claude Code or Codex wants agents to search and manage skills through native MCP tools.
  • A group evaluating skill quality wants to contribute reproducible results or publish a skill to the registry.
  • A team without access to the hosted registry wants to try search, recommendations, and installs against a local catalog.

What you need

Detected in the repository

  • Rust (from Cargo.toml)
  • Automated checks on GitHub Actions

License in plain words

MITpermissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the copyright and license notice
  • Share your changes: no

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Install using the project installer, then search the registry:

curl -fsSL skillrank.dev | sh
skillrank search playwright

For source builds, use stable Rust and run cargo build --release. See the README for setup, evaluation, local registry, and configuration details.

Alternatives

  • clawhub: ClawHub is a skill registry with publishing and vector search, while SkillRank also evaluates skill effects on token use, speed, and success.
  • skill-up: skill-up focuses on repeatable tests and regression reports for skills and workspaces, while SkillRank pairs evaluations with a public skill registry.
  • SkillOpt: SkillOpt uses evaluated task trajectories to improve skill documents through gated edits, rather than offering SkillRank's registry and paired skill-effect evaluations.
ProjectLanguageLicenseStarsStatus
SkillRankRustMIT0New
clawhubTypeScriptMIT9.5kActive
skill-upGoApache-2.01.4kNew
SkillOptPythonMIT18kNew

Considerations

  • The project is less than six months old and is changing quickly. It has recent commits and releases, four contributors, and no measured contributor concentration, but its short history still limits what can be judged about long-term stability.
  • The supplied requirements identify Rust and CI, but no tests are detected, even though the README documents cargo test. Verify the test setup and behavior for your intended use before adopting it in a critical workflow.
  • Evaluation results depend on the agent, suite, and run conditions. The project marks low trial counts as directional and limits non-Docker or non-reference-agent results to self-reported status.
  • The CLI and core library are MIT licensed, while the registry service and website use Elastic License 2.0, which restricts offering them as a competing hosted or managed service.
  • The hosted registry is a separate service. A local registry is available, but teams should check whether its seed or custom catalog meets their needs.

Found this useful?

Share it with someone who would like SkillRank.

Comparisons