JSON parse with prototype poisoning protection

JSON parse with prototype poisoning protection

Set of utilities to ease detection and revert of poisoning

Educational demo: a deliberately vulnerable npm package showing how GitHub Actions cache poisoning can produce a malicious release without stealing any credential. Do NOT use in production.

Production-grade CLI for auditing AI prompts. Detects invisible characters, homoglyph attacks, and adversarial Unicode poisoning in your development workflow.

Cache-Poisoning applied to Node.js require statements so you inject your own modules

Open-source agent security framework. Detects and defends against AI Agent Traps - content injection, embedded jailbreaks, RAG poisoning, data exfiltration, and more.

JSON parse with prototype poisoning protection

Open detection standard -- like Sigma, but for AI agents. Executable rules for prompt injection, tool poisoning, context exfiltration, and MCP attacks. Shipped in Cisco AI Defense. MIT-licensed.

Security scanner for MCP (Model Context Protocol) servers and configs — tool poisoning, secret leaks, rug-pull detection, excessive permissions, insecure transport.

> Like a fine steak, some dependencies are better aged. Unlike a fine steak, malicious ones won't just give you food poisoning.

BunnyHijack PoC - PATH poisoning demo against Bun trustedDependencies

MCP security guard and package manager: block prompt injection, tool poisoning and rug-pulls in Model Context Protocol servers at runtime; trust-scored installs for Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Gemini CLI.

MCP security scanner: scan MCP servers and tool manifests for tool poisoning, prompt injection, tool-name shadowing, excessive-agency schemas, unpinned/remote sources, and rug-pulls - 10 OWASP-mapped heuristics, CLI + MCP server, fully offline. Also runs

Security and cost audit for your AI agent toolchain — MCP servers, hooks, skills. One command finds poisoning, dangerous capabilities, context bloat and drift, and gates CI.

Name holder for security purposes, prevents registry poisoning attack

Deterministic Bounded-Context, Structural File Catalog & Anti-Poisoning Partition for AI Agents

29 security rules for MCP servers & agents — tool poisoning, prompt injection, exfiltration, attack chains, stale-trust caching. Ships semgrep ruleset + zero-dependency lite scanner.

A transparent audit proxy for MCP servers: records every tool call, detects tool poisoning and rug pulls

AI security scanner and runtime enforcement for LLM applications — detects prompt injection, API key leaks, unsafe output, RAG poisoning, and agent hijacking in code and live traffic.