agent-security-harness

---
name: agent-security-harness
description: >-
  Security test AI agent systems against protocol-level attacks.
  Use when: (1) testing MCP servers for tool poisoning, capability escalation, or protocol downgrade,
  (2) testing A2A agents for card spoofing, task hijacking, or context leakage,
  (3) testing L402/x402 payment flows for authorization bypass or receipt manipulation,
  (4) running pre-deployment adversarial testing for AIUC-1 certification readiness (B001, C010, D004),
  (5) enterprise platform security testing across 20 platforms,
  (6) full APT simulation with GTG-1002,
  (7) false positive rate testing (over-refusal),
  (8) supply chain provenance and attestation testing (CVE-2026-25253),
  (9) jailbreak resistance testing (DAN, token smuggling, authority impersonation).
  NOT for: model-layer testing (use Garak), identity/access policy enforcement (use MS Agent Governance),
  or static code scanning.
homepage: https://github.com/msaleme/red-team-blue-team-agent-fabric
metadata: {"openclaw":{"emoji":"🛡️","requires":{"bins":["python3","pip","agent-security"],"python":">=3.10"},"tags":["security","testing","mcp","a2a","l402","x402","agent-security","nist","owasp","red-team","aiuc-1","pre-certification","provenance","jailbreak","over-refusal"]}}
---

# Agent Security Harness

**332 security tests** across **24 modules** for AI agent systems. 4 wire protocols (MCP, A2A, L402, x402), 20+ enterprise platforms, GTG-1002 APT simulation, false positive rate testing, supply chain provenance, jailbreak resistance, AIUC-1 certification prep. Zero external dependencies for core protocol modules.

**Current version: v3.8.1** | [PyPI](https://pypi.org/project/agent-security-harness/) | [GitHub](https://github.com/msaleme/red-team-blue-team-agent-fabric) | Apache 2.0

**New in v3.8.1:** MCP Server (expose harness as MCP tools for any AI agent), Attestation Registry (opt-in, Ed25519 signed), Telemetry (opt-in, GDPR compliant), GitHub Action for CI/CD, Free MCP Secu