Model or dataset
tech-leads-club/agent-skills avatar
tech-leads-club/agent-skills

Agent Skills ships scanned SKILL.md bundles to nineteen coding agents

The secure, validated skill registry for professional AI coding agents. Extend Antigravity, Claude Code, Cursor, Copilot and more with absolute confidence.

7,167 stars570 forksTypeScriptNOASSERTION

At a glance

What is it?
tech-leads-club/agent-skills is an npm registry and CLI that copies packaged SKILL.md instructions into a project for Claude Code, Cursor, Copilot and sixteen other agents. Its security pipeline is spelled out in the README and SECURITY.md.
Who is it for?
Adopt the registry when you want a curated set of SKILL.md bundles and can accept a Node 24 floor plus a per project npx step. If you need per agent install flags, a rollback, or a catalog version you can pin, check the README's Quick Start and How It Works sections first.
Can I use it commercially?
Check first. The repository uses a licence we do not classify automatically, so read its LICENSE file before any commercial use.
Is it still maintained?
Yes. The repository last received commits 4 days ago.
What is it written in?
Mainly TypeScript, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • tech-leads-club/agent-skills7,167 ★ · +1,817
Star history
Repository2026-09-122026-11-15
tech-leads-club/agent-skills5.4k7.2k

1,817 Stars in the last 64 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

One npx command writes skills into your project

The whole install path is a single line:

bash
npx @tech-leads-club/agent-skills

Because it runs through npx, nothing is installed globally and there is no separate binary to upgrade later. The monorepo root declares its publish target as registry.npmjs.org with public access, and puts a floor under the runtime: `engines.node` is `>=24`. That floor is the first thing to check on a machine still on Node 20. The repository tree also carries `.claude-plugin/` and `.cursor-plugin/` directories, so one catalog serves more than one agent format.

SKILL.md is the unit of distribution, not a package

A skill here is a directory with a fixed shape:

code
packages/skills-catalog/skills/
  (category-name)/
    skill/
      SKILL.md          ← Main instructions
      templates/        ← File templates
      references/       ← On-demand documentation

`SKILL.md` carries the instructions, `templates/` carries files an agent can copy, and `references/` carries documentation pulled in only when needed. Categories are parenthesised directory names, and the featured entries name five of them: `(development)` for `tlc-spec-driven`, `(cloud)` for `aws-advisor`, `(web-automation)` for `playwright-skill`, `(design)` for `figma`, `(security)` for `security-best-practices`. The split matters because a long reference document does not have to sit inside the main instruction file.

The scan step sits between a catalog tag and a publish

The catalog is versioned separately from the code that installs it. Three scripts in the root `package.json` describe that path:

bash
npx nx run @tech-leads-club/skills-catalog:security-scan
bash
npm run scan && nx release

`prepublishOnly` is `npm run scan`, so a publish cannot skip it. The scanner named in the security notes is Snyk Agent Scan, called mcp-scan before the rename, run over every skill before publication. What comes out the other side are separate tags for the catalog, the MCP server and the main package: `skills-catalog-v0.17.19`, `mcp-v0.1.11` and `v1.6.3`, all published on 2026-11-02.

Symlink guards and an atomic lockfile, with no documented way back

The hardening list is specific: sanitisation, path isolation, symlink guards, an atomic lockfile, an audit trail, lockfiles plus content hashing for immutability, static analysis in CI, human curated prompts, and a catalog with no binaries. The threat model and the reporting process live in `SECURITY.md`. The README justifies the approach by pointing at a Snyk report on open marketplaces and the 13.4% critical issue figure quoted there, which is an argument about other registries rather than a measurement of this one.

Eighteen agents in three tiers, reached by the same command

Support is announced as a table. Tier 1 holds Claude Code, Cline, Cursor, GitHub Copilot and Windsurf. Tier 2 holds Aider, Antigravity, Gemini CLI, Kilo Code, Kiro, OpenAI Codex, Roo Code and TRAE. Tier 3 holds Amazon Q, Augment, Droid from Factory.ai, OpenCode, Sourcegraph Cody and Tabnine. The README's Quick Start installs skills with one npx invocation.

The README also has an MCP Server section, and the MCP server is a separate workspace package exposed in the dev scripts as `@tech-leads-club/agent-skills-mcp` through `npm run start:dev:mcp`.

An Nx workspace hides behind a private 0.0.0 root version

The monorepo only matters if you want to contribute or vendor a skill. Its root `package.json` is named `@tech-leads-club/agent-skills-monorepo`, sits at version `0.0.0`, sets `private` to true and `type` to `module`, and declares workspaces of `packages/*` and `libs/*`. New skills are scaffolded through a generator plugin:

bash
nx g @tech-leads-club/skill-plugin:skill

Validation runs over the catalog directory as a batch:

bash
npx tsx tools/validate-skills.ts --batch packages/skills-catalog/skills

For someone who only installs skills, the Nx layer, the Jest configuration and the ESLint config are all bypassed.

Updating, pinning and offline installs are undocumented

Two gaps decide whether this registry fits your workflow. First, check the README's Quick Start and How It Works sections and `CHANGELOG.md` for how updates, pinning and offline installs are handled. Second, licensing needs a decision from you rather than a fact from the metadata: the root `package.json` declares MIT and marks the workspace private, the repository license field records no identifier, and a `LICENSE` file exists. The README reserves a section for licence and attribution, so read that text before a bundled skill ships inside your product. If your process needs an audit trail of what each project received, the lockfile is the artefact.

Editorial conclusion

Adopt the registry when you want a curated set of SKILL.md bundles and can accept a Node 24 floor plus a per project npx step. If you need per agent install flags, a rollback, or a catalog version you can pin, check the README's Quick Start and How It Works sections first. Verify first what the LICENSE file covers for the skills you copy, because the repository license field records no identifier and the workspace root is marked private.

Frequently asked questions

How do I install agent skills in Claude Code?

Run `npx @tech-leads-club/agent-skills` from the project that should receive the skills. Claude Code is listed in Tier 1 of the supported agents, and the package requires Node 24 or newer.

How do I use agent skills in Cursor?

Cursor is a Tier 1 target and takes the same install command as every other agent. The repository also ships a `.cursor-plugin/` directory next to `.claude-plugin/`.

How do I use agent skills?

Skills are directories under `packages/skills-catalog/skills/(category-name)/skill/`, each holding `SKILL.md` plus `templates/` and `references/` folders, grouped into parenthesised categories such as `(development)`, `(cloud)` and `(security)`. Installation happens once per project with `npx @tech-leads-club/agent-skills`.

How do I use agent skills in OpenCode?

OpenCode appears in Tier 3 (Enterprise) alongside Amazon Q, Augment, Droid, Sourcegraph Cody and Tabnine. An OpenCode user runs the same `npx @tech-leads-club/agent-skills` command as a Claude Code user.

Official sources

  1. Issues
  2. Project website
  3. README
  4. Releases
  5. tech-leads-club/agent-skills on GitHub
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/tech-leads-club-agent-skills.svg)](https://hysenlabs.com/projects/tech-leads-club-agent-skills)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/tech-leads-club-agent-skills.svg)](https://hysenlabs.com/projects/tech-leads-club-agent-skills)

Was this useful?