#agent-security

Open source authorization engine for AI agents. Confidence-aware gating · Human-in-the-loop review · Policy-as-code · Full audit trail

Open-source security-first runtime for AI agents that learn and act across sessions.

AI agent security harness, adversarial testing: 623 executable tests in the v4.21.3 release, across MCP, A2A, x402/L402, decision governance, benchmark integrity, human-in-the-loop, skill supply chain. Four target shapes: a verdict must be able to be wrong AND to be right. OWASP Agentic v1.1 T1-T17, AIUC-1 19/20 (2026-Q1/Q2 set), NIST AI 800-2.

The agent framework where the model never holds the trigger — every consequential action clears your policy first, waits for a human when it matters, and lands on a record you can verify. Build on it, or put it around the agent you already have.

MCP servers expose tools with no information about what they actually do at runtime. mcpsafetywarden sits between your agent and any MCP server, profiling tool behavior, blocking destructive calls, and running active security audits before you trust them in a workflow.

Local credential control for AI coding agents.

Runtime policy enforcement and audit control plane for MCP tool execution. Deterministic, non-AI policy engine that intercepts MCP tools/call requests before execution.

BonkLM - LLM Security Guardrails with Interactive Setup Wizard

Security & Observability for AI Agents — audit every tool call, enforce allow/block policies, catch prompt injection and data leaks. Local-first; works with Claude Code, Codex, Copilot CLI, OpenClaw, LangChain, and all major LLM APIs.

Python SDK for Agent Passport System identity, scoped delegation and signed receipts.

Policy decision runtime for agent security, built on the agent-hooks control contract.

Open source input firewall for AI agents, beta. A local scanner checks text, code, PDFs, images, QR codes, audio and video with 1,583 patterns across 117 categories and reports findings and incomplete scans. A Claude Code hook blocks secret material in tool calls and the paths and hosts your policy lists before tools run.

Agentic AI EDR for developer workstations and autonomous agent swarms. Build Swarm Detection & Response platforms with Clawdstrike.

Web4: an open standard for verifiable AI presence — LCT identity, T3/V3 trust tensors, MRH context, ATP/ADP. Rust + Python SDKs on crates.io/PyPI. AGPL-3.0-or-later.

Deterministic governance for Claude and MCP tool calls. Pin approved capabilities, detect drift, and refuse unsafe or unapproved actions before execution. No model in the decision path.

Static security scanner for AI agents, MCP tools, and LLM applications. Detects OWASP LLM Top 10 risks, unsafe tool use, prompt injection paths, secrets, and AI infrastructure misconfigurations. Zero runtime dependencies.

Zero-dependency Web Crypto suite & Model Context Protocol (MCP) server for AES-256-GCM, RSA-4096, and AI agent security.

Audit your AI agent configs before you run them — prompt injection, rogue MCP servers, credential-exfil. Plus dependency vetting.

Local-first trust layer for AI coding agents: structural prompt-injection and action guards, secret scanning, semantic memory, code review, and 17 MCP tools. No model API required.

OpenAIDR — session collection for AI coding agents. Reads what agents write to disk and normalises it into one session model.

Evidence-backed AppSec for AI applications and agents. Powered by HAIEC.