Technical Documentation

A comprehensive deep dive into the architecture, detection logic, and internal workflows of the skillstack CLI.

────────────────────────────────────────────────

[01]

System Architecture

CLI Layer

Orchestrates the entire lifecycle from command parsing to final installation. It manages the asynchronous state machine that coordinates between detectors and the remote registry.

Core Engine

Consists of modular detectors that perform static analysis on project manifests, configuration files, and directory structures to identify technology signatures.

→Technology DetectorAnalyzes package.json, Cargo.toml, and build.gradle.
→Combo DetectorIdentifies stacks like 'Next.js + Tailwind + Supabase'.
→Agent DetectorDetects installed AI agents like Claude Code or Cursor.

[02]

Execution Workflow

Scan

Recursive directory traversal to locate project workspaces and configuration manifests.

Analyze

Pattern matching against the internal technology-to-skill mapping registry.

Fetch

Retrieval of skill metadata and installation scripts from the skills.sh API.

Apply

Atomic file system operations to inject skills into agent-specific dot-directories.

[03]

Monorepo Structure

skillstack/
├── packages/
│   ├── skillstack/          # CLI core & orchestrator
│   │   ├── installer.ts     # FS operation logic
│   │   └── registry.ts      # API communication
│   └── autoskills/          # Signature matching patterns
├── src/                     # Documentation portal
└── assets/                  # Core branding assets

[04]

Configuration

Example .skillstackrc.json

{
  "exclude": ["legacy-patterns"],
  "include": ["experimental-react-19"],
  "target": "claude-code",
  "autoConfirm": true
}

────────────────────────────────────────────────