Skip to content
Nexload SDK

Install agent skills

The repository ships validated, self-contained Skills. The official skills CLI discovers the current skills/**/SKILL.md files, so the list command is the source of truth instead of a hard-coded count. Each Skill keeps its core workflow in SKILL.md, routes deeper package contracts through references/, and commits behavior plus trigger cases under evals/.

  1. Inspect the available Skills before installing:

    Terminal window
    npx skills add gecut/nexload-sdk --list

  2. Install a specific Skill for the current project:

    Terminal window
    npx skills add gecut/nexload-sdk --skill healthcheck-core

  3. Update installed Skills when you intentionally want newer guidance:

    Terminal window
    npx skills update

Example: a Next.js health route

Section titled “Example: a Next.js health route”
Terminal window
npx skills add gecut/nexload-sdk --skill healthcheck-nextjs-routes

Source layout

Section titled “Source layout”
  • Directoryskills
    • Directoryhealthcheck
      • Directorynextjs-routes
        • SKILL.md
        • Directoryreferences
          • route-matrix.md
          • protection-and-proxy.md
          • app-router-playbook.md
        • Directoryevals
          • evals.json
          • trigger-evals.json
    • Directorypayload-fields
      • Directorymoney
        • SKILL.md
        • references
        • evals
    • Directorypayload-editor
      • Directorycore
        • SKILL.md
        • references
        • evals
    • Directorypayload-schema
      • Directoryuse
        • SKILL.md
        • references
        • evals
      • Directorydevelop
        • SKILL.md
        • references
        • evals
    • Directorypayload-operations
      • Directorycore
        • SKILL.md
        • references
        • evals
      • Directoryclient
        • SKILL.md
        • references
        • evals
      • Directoryserver
        • SKILL.md
        • references
        • evals

Run pnpm skills:validate in a repository checkout to verify frontmatter, path/name consistency, references, line limits, behavior evals, and balanced trigger evals.