vulntext 2025.12.21084032

A new package that analyzes user-provided text to extract structured information about software security vulnerabilities. It processes natural language descriptions of code issues or security concerns

vulntext - Vulnerability Text Analyzer

vulntext is a lightweight Python package that extracts structured vulnerability information from natural‑language descriptions of software security issues. By feeding the package a text (e.g., a bug report, CVE description, or a security advisory) it returns an array of structured data items such as vulnerability type, severity level, affected component, and recommended remediation steps.

Features

• Simple, self‑contained interface
• Uses LLM7 by default (free tier, suitable for most uses)
• Fully flexible: plug in any LangChain compatible LLM
• Generates results that match a user‑defined regex pattern

Installation

pip install vulntext

Quick Start

from vulntext import vulntext

user_input = """
A buffer overflow bug in the network packet parser allows an attacker to crash
the service and potentially execute arbitrary code. The vulnerability is
present in version 2.3.4 of the packet-processor library.
"""

# Basic usage (uses the default LLM7 wrapper)
results = vulntext(user_input)
print(results)

You will see an output that looks roughly like:

[
  {
    "type": "Buffer Overflow",
    "severity": "High",
    "component": "packet-processor",
    "version": "2.3.4",
    "remediation": "Update to 2.3.5 or patch the parser."
  }
]

Using a Custom LLM

vulntext accepts a LangChain BaseChatModel instance. This allows you to use any provider supported by LangChain:

# OpenAI
from langchain_openai import ChatOpenAI
from vulntext import vulntext

llm = ChatOpenAI()
response = vulntext(user_input, llm=llm)

# Anthropic
from langchain_anthropic import ChatAnthropic
llm = ChatAnthropic()
response = vulntext(user_input, llm=llm)

# Google Generative AI
from langchain_google_genai import ChatGoogleGenerativeAI
llm = ChatGoogleGenerativeAI()
response = vulntext(user_input, llm=llm)

When a custom LLM is provided, the api_key argument is ignored because the key is managed by the wrapper you pass in.

Optional API Key for LLM7

If you want to override the free LLM7 tier or need higher rate limits, supply your own key either as an environment variable or directly:

export LLM7_API_KEY=your_basic_api_key_here

or

response = vulntext(user_input, api_key="your_basic_api_key_here")

You can obtain a free key by registering at LLM7.

Parameters

Parameter	Type	Description
user_input	str	The text containing the vulnerability description.
llm	Optional[BaseChatModel]	A LangChain chat model instance. When omitted, the package falls back to the bundled ChatLLM7.
api_key	Optional[str]	Your LLM7 API key. Ignored if llm is supplied. Note that the free tier is usually sufficient for most uses.

Development & Bug Reports

• Repository: https://github.com/chigwell/vulntext
• Issues: https://github.com/chigwell/vulntext/issues

License

MIT

Author

Eugene Evstafev
📧 [e-mail]
🐙 @chigwell

---

Happy hacking! 🎯