* [Pipelock v3.4.0](https://github.com/luckyPipewrench/pipelock) – All-in-one security harness for AI agents offering egress control, DLP, SSRF protection, prompt-injection scanning, and workspace integrity monitoring.

* [ToolHive v0.44.0](https://github.com/stacklok/toolhive) – Simplifies secure deployment and management of Model Context Protocol servers in isolated containers.

* [Arcjet v1.11.0](https://github.com/arcjet/arcjet-js) – SDK for implementing security features like bot detection, rate limiting, and email validation in JavaScript applications.

* [AgentHound 1.1.0](https://github.com/adithyan-ak/AgentHound) – Offensive security framework for AI agent infrastructure that runs recon, credential and model inventory, exploitation, and builds a Neo4j graph to analyze attack paths.

* [Rampart v1.8.0](https://github.com/peg/rampart) – Security layer enforcing YAML policies for commands, file access, and network requests from AI coding agents.

* [AgentScan v1.1.0](https://github.com/7anX/AgentScan) – HTTP-based scanner that discovers exposed MCP servers, A2A Agent Cards, and open LLM APIs using one command-line workflow.

* [Codex Security npm-v0.1.19](https://github.com/openai/codex-security) – A CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities in your code.

* [Julius v1.4.12](https://github.com/praetorian-inc/julius) – LLM service fingerprinting tool that identifies which AI server software runs on network endpoints in seconds.

* [OPFOR v0.10.2](https://github.com/KeyValueSoftwareSystems/agent-opfor) – Adversary emulation tool that red-teams AI agents and MCP servers end to end across OWASP-style test suites with logged, judge-evaluated results.

* [Augustus v0.14.15](https://github.com/praetorian-inc/augustus) – Go-based LLM vulnerability scanner testing models for prompt injection, jailbreaks, encoding exploits, and data extraction.

* [AEGIS aegis-v0.13.0-alpha](https://github.com/antropos17/Aegis) – Local-first EDR-style monitor tracking AI agent processes, file and network activity with per-agent risk scoring.

* [Hivemind 2026.08.07-rc](https://github.com/hivementality-ai/hivemind) – Self-hosted multi-agent AI team platform with persistent workspaces, team chat, sandboxed tool execution, and integrated messaging channels.