Skip to content
Projects

miunasu/IDA-Skill has 238 GitHub stars and 39 forks. Star changes are not measured yet. Last checked 2026-11-15 23:40 UTC. Last push 2026-10-21.

How these counts are measured

Measured star growth

in 24h
Update delayed
in 7 days
Update delayed
in 15 days
Update delayed
in 1 month
Update delayed

Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.

About

使用skill让 AI Agent 像安全分析师一样分析恶意样本 | AI Agent skill for automated malware analysis using IDA Pro

AI

README_EN.md

Original

IDA Skill for AI Agent

English | 中文

🤖 Enable AI Agents to analyze malware like security analysts

What is this?

This is an AI Agent skill that enables AI to automatically analyze malware using IDA Pro, just like human security analysts:

  • 🔍 Automatic Malicious Behavior Detection - Analyze C2 communication, process injection, persistence mechanisms, and malicious behaviors
  • 🧩 Understand Code Functionality - Automatically analyze function logic, identify encryption algorithms, and trace data flows
  • 📊 Extract Key Information - Import tables, strings, opcode features, YARA rules
  • 🎯 Locate Threat Indicators - IP addresses, domains, file paths, registry keys

Demo Video

Demonstration using my multi-agent framework Spore: Since the video embedded in the readme file failed to upload, please download movie.mp4.

How to Use?

Super simple! Just 3 steps:

1️⃣ Configure IDA Path

Edit config.json:

{
  "idat_path": "C:/Program Files/IDA Pro 9.0/idat64.exe"
}

2️⃣ Let AI Agent Use the Skill

Simply place the IDA Skill in your skill folder.

3️⃣ Start Analysis

AI Agent will automatically:

  • Initialize IDA database
  • Extract import tables, strings, decompiled code
  • Perform reverse analysis like a human
  • Analyze functionality and extract C2
  • Identify encryption algorithms and suspicious behaviors
  • Generate analysis reports

That's it! 🎉

Built-in Tools

This skill package integrates multiple powerful analysis tools that the Agent will automatically invoke:

🤖 REAI - AI Function Analysis

Based on my open-source project REAI

Features: Use LLM to automatically understand function functionality and recursively analyze call chains

  • Automatically identify function purposes (e.g., "decrypt config", "connect to C2")
  • Automatically rename functions (sub_401000 → AI_decrypt_config)
  • Add comments at call sites explaining sub-function purposes

Configuration: Edit tools/reai.py to set your LLM API

API_KEY = "sk-..."           # Your API key
API_URL = "https://..."      # API endpoint (supports OpenAI/Azure/DeepSeek/local models)
MODEL = "gpt-4"              # Model name

🔍 FindCrypt - Cryptographic Algorithm Detection

Based on the open-source FindCrypt project

Features: Automatically detect cryptographic constants in code

  • Supported algorithms: AES, DES, RC4, Blowfish, TEA
  • Supported hashes: MD5, SHA1, SHA256
  • Supported encodings: Base64, CRC32

📊 Export Check - Export Table Analysis

Features: Detect abnormal DLL export functions

  • Analyze export function sizes
  • Identify abnormal export patterns

FAQ

Q: Do I need to know IDAPython?

A: No! Just let the AI Agent read SKILL.md, and the Agent will automatically use these tools. You only need to describe your requirements in natural language.

Q: Can I use it commercially?

A: No. This project is licensed under GPL-3.0 for educational and research purposes only. Commercial use is prohibited. Please contact the author for commercial licensing.

License

GPL-3.0 License - See LICENSE file for details

This project is licensed under GPL-3.0, for educational and research purposes only, commercial use is prohibited.

  • ✅ Free to use, modify, and distribute
  • ✅ Must remain open-source, derivative works must also use GPL-3.0
  • ❌ Commercial use prohibited
  • ❌ Closed-source use prohibited

Disclaimer: When using this project for reverse engineering, you must comply with all applicable laws and regulations, including software license agreements. The author is not responsible for any consequences arising from the use of this project.