The AI Agent Security Lifecycle

Nine open-source tools. One security lifecycle. Pick the stage you're in.

All tools

Each tool works standalone or through the OpenA2A CLI unified interface.

OpenA2A CLI

Build

Unified entry point that orchestrates all security tools through adapter commands

WHODeveloper / DevOps / Security
HOWCLI
$ npx opena2a-cli init
Trust score 0-100Scope drift detectionConfig signingShadow AI detection

HackMyAgent

Scan

147 security checks, 55 adversarial payloads, auto-fix with rollback, OASB compliance

WHOSecurity / Developer
HOWCLI / Library
$ npx hackmyagent secure
147 checks / 30 categories55 attack payloadsAuto-fix + rollbackCIS-style reports

Secretless AI

Protect

Keeps credentials out of AI context windows across Claude Code, Cursor, Copilot, and more

WHODeveloper / DevOps
HOWCLI / Hook
$ npx secretless-ai init
49 credential patternsMulti-backend storageMCP server encryptionDLP transcript scanning

AIM

Build

Cryptographic identity, capability policies, trust scoring, and audit logging for AI agents

WHOEnterprise / DevOps / Security
HOWDocker / SDK / API
$ docker compose up
Ed25519 identityPolicy-as-code8-factor trust scoreMCP attestation

OASB

Scan

222 standardized attack scenarios across 8 vulnerability categories with compliance scoring

WHOSecurity / Compliance
HOWCLI / Library
$ npx opena2a-cli benchmark
222 test scenarios8 attack categories3 maturity levelsCompliance reports

ARP

Monitor

Runtime process, network, and filesystem monitoring with protocol-aware AI endpoint detection

WHODevOps / Security
HOWCLI
$ npx opena2a-cli runtime start
Process monitoringNetwork inspectionMCP/A2A/OpenAI detectionFilesystem watching

AI Browser Guard

Protect

Chrome extension that detects, monitors, and controls AI agents operating in your browser

WHOEnd User / Security
HOWChrome Extension
$ Install from Chrome Web Store
3-layer agent detectionEmergency kill switchDelegation wizardSession timeline

DVAA

Train

Deliberately vulnerable AI agent platform with 10 agents, 18 CTF challenges, and 3 protocol types

WHOSecurity / Trainer
HOWDocker
$ docker pull opena2a/dvaa
10 vulnerable agents18 CTF challengesMCP + A2A + OpenAIWeb dashboard

ai-trust

Verify

Trust verification CLI for AI packages. Queries the OpenA2A Registry for security scans, dependency risk, and known advisories.

WHODeveloper / DevOps
HOWCLI
$ npx ai-trust check <package>
Single-package trust lookupDependency auditBatch verification5 trust levelsScan on demandCommunity contributions

Coverage matrix

Map your threat model to the right tools. Every row is covered by at least two tools.

ThreatCLIHMASecretlessAIMOASBARPai-trustGuardDVAA
Hardcoded credentials
Prompt injection
MCP config tampering
Scope drift
Unauthorized agent access
Supply chain risk
Data exfiltration
Privilege escalation

How they connect

All tools work standalone. The CLI orchestrates them together.