Wide SDK

Python SDK to build autonomous AI coding and web security agents powered by Wide

Download · Build · Changelog · Wide IDE · Wide CLI · Wide App · Wide Action · Wide Rules · License

Overview

Wide SDK allows Python developers to programmatically build, customize, and orchestrate autonomous AI coding and penetration testing agents with native Wide security tooling.

• Autonomous Agent Loop: Full ReAct (Reasoning and Action) loop with dynamic context compaction and execution limits.
• Native SAST Engine: Programmatic access to Wide's 20+ vulnerability inspection rules (SQLi, XSS, SSRF, Command Injection, XXE, Deserialization, Hardcoded Secrets).
• Security Fuzzing and Payloads: Built-in web endpoint fuzzer and attack vector generator for automated verification.
• Multi-Provider Architecture: Pluggable LLM backends supporting Ollama (local offline), Google Gemini, Anthropic Claude, and OpenAI/DeepSeek.
• Custom Tool Ecosystem: Extensible @tool interface with automatic JSON Schema generation from Python function signatures and type hints.

Installation

pip install wide-sdk

Quick Start

from wide_sdk import Agent, GeminiProvider

provider = GeminiProvider(api_key="YOUR_GEMINI_API_KEY")
agent = Agent(provider=provider, working_directory=".")

response = agent.run("Scan this project for SQL injection vulnerabilities and fix any confirmed issues.")
print(response)

Security Scanner API

Run Wide's comprehensive static analysis engine directly in Python:

from wide_sdk.security import SecurityScanner

scanner = SecurityScanner(workspace_dir=".")
results = scanner.scan(min_severity="HIGH")

for finding in results.findings:
    print(f"[{finding.severity}] {finding.category} in {finding.file}:{finding.line}")
    print(f"Issue: {finding.message}")
    print(f"Remediation: {finding.remediation}\n")

Active Fuzzer API

Automatically probe an API endpoint parameter:

from wide_sdk.security import fuzz_endpoint

report = fuzz_endpoint(
    url="http://localhost:3000/api/search",
    param="q",
    payload_type="sqli"
)

if report.is_vulnerable:
    print(f"Vulnerability confirmed on {report.param}!")

Configuration

Set up environment variables for your preferred provider:

export GEMINI_API_KEY="your-gemini-key"
export ANTHROPIC_API_KEY="your-claude-key"
export OPENAI_API_KEY="your-openai-key"
export DEEPSEEK_API_KEY="your-deepseek-key"

License

GPL-3.0-or-later sl4de