agent-ecosystem

Agent Skill Implementation — Development skill for Claude Code

Development community

Empirical research into how agent platforms implement Agent Skill loading, management, and presentation.

How to install Agent Skill Implementation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open agent-ecosystem/agent-skill-implementation and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agent Skill Implementation does

Empirical research into how agent platforms implement Agent Skill loading, management, and presentation. Benchmark skills and per-platform findings.

Alternatives in Development

OpenAI Codex CLI — (Rust implementation) 67.8k ★
Research Codebase — You are tasked with conducting comprehensive research across the codebase to answer user questions by spawning 10k ★
Financial Services Plugins — Plugins that turn Claude into a specialist for financial services — investment banking, equity research, priva 7.7k ★

README

Agent Skill Implementation

[Agent Skills](https://agentskills.io) are markdown files and related resources that help AI agents perform tasks. The spec defines a portable format: write a skill once, and users can install it on any platform that supports the specification. Over 25 platforms currently implement some level of Agent Skill support.

But "some level" is doing a lot of work in that sentence. If you're publishing skills for others to use, your users could be on any of those 25+ platforms, and each one may load, present, and manage your skill differently. A skill that works perfectly on the platform you tested it on may lose access to its reference files, have its metadata stripped, or find its instructions pruned from context on another, with no error to tell you any of it happened.

This project investigates how skill implementation actually works in practice, from two angles: how **platforms** load and manage skills, and how **authors** write them. The platform side is the current focus. For the author side, see [Related research](#related-research) below.

Platform implementation research

The spec recommends a "progressive disclosure" loading model, but gives platforms wide latitude in implementation. The [client implementation guide](https://agentskills.io/client-implementation/adding-skills-support) provides more detailed guidance, but was derived from analysis of 7 of those 25+ platforms and published months after most had already shipped their implementations. This project exists to find out what platforms actually do, through empirical testing rather than assumptions.

What's here

**[checks.md](checks.md)**: 46 checks across 10 categories of platform behavior (loading, validation, execution, and access control) that need empirical testing. Each check describes what it evaluates and why it matters for skill authors.
**[benchmark-skills/](benchmark-skills/)**: 45 spec-compliant skills designed to exercise those checks. Each cont