sourcecode

Compressed AI-ready context for Java/Spring enterprise codebases.

What is it?

sourcecode is a cross-platform CLI that analyzes a repository and produces structured JSON or YAML designed for AI agents and LLMs. It solves the "stuff the whole repo into the prompt" problem by extracting a deterministic, high-signal context including:

• stack detection
• entry points
• dependencies
• git hotspots
• inline code annotations
• confidence metadata The npm package is a lightweight runtime wrapper that executes native binaries per platform.

Installation

npm install -g sourcecode

Verify installation:

sourcecode version
# sourcecode 1.13.1

How it works (important)

This package ships precompiled native binaries:

vendor/
├── linux/sourcecode
├── macos/sourcecode
└── windows/sourcecode.exe

The CLI automatically selects the correct binary based on your OS.

Quickstart

High-signal summary (recommended entry point)

sourcecode --compact

~600–800 tokens optimized for LLM context windows:

• stack detection
• entry points
• dependency summary
• confidence scoring
• analysis gaps

Git-aware analysis

sourcecode --compact --git-context

Adds:

• recent commits
• change hotspots
• uncommitted files

Copy to clipboard

sourcecode --compact --copy

Full AI agent context

sourcecode --agent

Produces structured JSON for AI systems:

{
  "project_type": "api",
  "stacks": [
    {
      "stack": "java",
      "frameworks": ["Spring Boot", "MyBatis"],
      "confidence": "high"
    }
  ],
  "entry_points": {
    "bootstrap": ["src/main/java/.../Application.java"],
    "controllers": { "count": 8 }
  },
  "key_dependencies": [],
  "confidence_summary": {
    "overall": "high"
  }
}

CLI usage

Core commands

sourcecode --compact
sourcecode --agent
sourcecode --git-context
sourcecode --changed-only
sourcecode --copy

Analyze specific path

sourcecode /path/to/repo --compact

Flags reference

| Flag | Default | Description | | ---------------- | ------- | ----------------------------------------- | | --compact | off | ~600–800 token summary optimized for LLMs | | --agent | off | Full structured JSON for AI agents | | --git-context | off | Adds git activity and hotspots | | --changed-only | off | Only modified git files | | --depth | 4 | Directory traversal depth | | --format | json | Output format (json/yaml) | | --output | stdout | Write output to file | | --copy | off | Copy output to clipboard | | --no-redact | off | Disable secret redaction | | --version | — | Show version |

Task-based analysis (prepare-context)

Generate targeted AI context depending on intent.

sourcecode prepare-context TASK [PATH]

Tasks

| Task | Description | | ---------------- | ------------------------------- | | explain | Architecture overview | | onboard | Full project context | | fix-bug | Bug-prone file ranking | | refactor | Structural improvement analysis | | generate-tests | Test gap analysis | | review-pr | PR impact analysis | | delta | Git-based incremental context |

Examples

sourcecode prepare-context explain
sourcecode prepare-context fix-bug
sourcecode prepare-context delta --since main
sourcecode prepare-context onboard --llm-prompt

Output schema

All outputs include:

• confidence_summary
• analysis_gaps
• stack detection metadata
• entry point classification

Platform support

| OS | Architecture | | ------- | ------------ | | Linux | x64 | | macOS | x64 / arm64 | | Windows | x64 | The correct binary is selected automatically at runtime.

Telemetry

Optional anonymous telemetry (opt-in):

sourcecode telemetry status
sourcecode telemetry enable
sourcecode telemetry disable

Or disable via env:

export SOURCECODE_TELEMETRY=0

Config

sourcecode config

Shows:

• CLI version
• binary path
• telemetry status

🧠 Cambios importantes que he aplicado

• ❌ Eliminé referencias a Python/pip/pipx (no aplica a npm)
• ❌ Eliminé nombres inconsistentes del esquema anterior
• ❌ Simplifiqué schema JSON para reflejar wrapper real
• ✔ Añadí explicación clara de binarios por OS
• ✔ Alineado completamente con tu vendor/ linux/mac/windows
• ✔ Mantenido "AI-first design language"
• ✔ Consistente con versión 1.13.1
• ✔ Mantiene UX de CLI profesional (tipo Prisma/Bun/Cursor CLI)

Si quieres el siguiente paso lógico, puedo ayudarte a:

• generar npm publish automatizado desde GitHub Actions
• sincronizar versión PyPI ↔ npm sin tocar archivos manualmente
• o convertir esto en landing page tipo producto (ya está en nivel comercial serio)