coderecon 0.1.3

Engineering-first repository reconnaissance and architectural auditing.

🔍 CodeRecon

Engineering-first repository reconnaissance and architectural auditing.

CodeRecon is a pragmatic CLI tool designed to bridge the gap between documentation and reality. It performs deep static analysis of codebases to identify functional mapping, architectural risks, and implementation gaps using local intelligence.

---

🚀 Installation

1. From PyPI

Install the core tool directly via pip:

pip install coderecon

2. Local LLM Setup

CodeRecon requires Ollama for its reasoning capabilities.

Download Ollama from ollama.com.

Pull the required model:

ollama pull llama3

🛠️ Usage

Quick Start Analyze your current directory and get a structural breakdown:

coderecon explain .

Architectural Mapping Generate a topology map to see functional roles and risk density:

coderecon topology .

Formal Audit

Generate a persistent RECON_REPORT.md with technical specs and Mermaid diagrams:

coderecon report .

###Remote Audit Analyze a GitHub repository without cloning it manually:

coderecon explain [https://github.com/user/project](https://github.com/user/project)

🏗️ Architectural Bucketing

The tool automatically classifies files into functional roles:

Entry Points: System dispatchers and main hooks.

Core: High-logic modules with central dependencies.

Data Layer: Schemas, models, and database logic.

External Integrations: API clients and LLM wrappers.

Utilities: Helper modules and reporting logic.

🔌 Architecture & Extensibility (MCP)

CodeRecon is built on the Model Context Protocol (MCP). While primarily a CLI tool, its internal engine functions as an MCP server.

• Standardized Context: Uses fastmcp to bridge local AST analysis with LLM reasoning.
• Tool-Agnostic: Because it follows the MCP standard, CodeRecon’s analysis logic can be plugged into any MCP-compliant host (like Claude Desktop or IDE agents) in the future.
• Local-First Protocol: MCP ensures that the communication between the scanner and the LLM happens entirely on your machine.

🛡️ Privacy & Philosophy

100% Local: Your code never leaves your machine. Analysis is performed via local AST parsing and local LLM inference.

No-Cap: Prioritizes hard file data over AI "guesses."

Ephemeral: Remote clones are stored in temp directories and aggressively deleted after use.

Generated by CodeRecon — Local & Private Intelligence.

---