Agent Debug Toolkit (ADT)

Find bugs in AI agent code before they find you.

Static analysis CLI that scans Python agent code for bugs, vulnerabilities, and performance issues. Zero dependencies — pure stdlib.

Free & Open Source

Analyzer	What it finds
Loop	Infinite loops, missing termination, missing error handling
Tools	Unsafe eval()/exec(), missing descriptions, missing error handling
Injection	Prompt injection via f-strings, missing delimiters
Memory	Unbounded collections, missing cleanup, global mutable state

Pro ($9 one-time → get license)

Analyzer	What it finds
RAG	Chunk size issues, low overlap, missing k-parameter
Performance	Token waste, expensive models, sequential API calls

Install

pip install git+https://github.com/iZhangDM/agent-debug-toolkit.git

Usage

adt analyze your_agent.py
adt analyze your_agent.py --severity critical --format json

Pro

Email 2638884823@qq.com. $9, lifetime license. You get:

• RAG pipeline analyzer
• Performance benchmark
• Priority support

License

Free tier: MIT. Pro features: proprietary.