Skip to main content

Getting started

Find the right tool for what you need to do.

Start here: Run review to get a full security report, then follow the recommendations.

npx opena2a-cli review

"I want to scan a project for security issues"

Start with the CLI. It runs a security assessment and gives you a trust score.

npx opena2a-cli init
opena2a init docs

"I want to check the trust score of an AI agent or MCP server"

Use trust to look up the trust profile, security posture, and supply chain data for any package.

opena2a trust @anthropic/mcp-server-fetch
opena2a trust docs

"I want to claim my agent and improve its trust score"

Use claim to verify your publisher identity and link it to your trust profile.

opena2a claim
opena2a claim docs

"I want to find vulnerabilities in my AI agent"

Use HackMyAgent for 310 static + 29 semantic + 164 adversarial checks and attack simulation.

opena2a scan --attack
HackMyAgent docs

"I have hardcoded API keys in my code"

Use protect to detect and migrate credentials to a secure vault.

opena2a protect
opena2a protect docs

"I want to prevent AI tools from leaking my credentials"

Use Secretless to block credential access in LLM contexts.

npx secretless-ai init
Secretless docs

"I want to monitor my agents at runtime"

Use ARP (Agent Runtime Protection) for process, network, and filesystem monitoring.

opena2a runtime start
ARP docs

"I want to manage agent identities and trust"

Use AIM for Ed25519 identity, trust scoring, MCP security, and ABOM.

docker compose up -d
AIM docs

"I want to benchmark my agent against a standard"

Use OASB for 222 standardized attack scenarios and compliance scoring.

opena2a benchmark
OASB docs

"I want to learn AI agent security hands-on"

Use DVAA - an intentionally vulnerable agent you can practice attacking.

docker run -p 3000:3000 opena2a/dvaa
DVAA docs

"I want to detect AI agents in the browser"

Use BrowserGuard - a Chrome extension with 4-layer detection.

BrowserGuard docs

Tools at a glance

ToolPurposeInstall
OpenA2A CLIUnified security CLInpm i -g opena2a-cli
HackMyAgentSecurity testing and attack simulationnpm i -g hackmyagent
Secretless AIKeep secrets out of AI toolsnpm i -g secretless-ai
AIMAgent identity and governancepip install aim-sdk
OASBAttack-scenario benchmark for agent security toolsTool page
Runtime ProtectionProcess, network, and filesystem monitoring for running agentsnpm i -g arp-guard
AI Browser GuardDetect and monitor browser AI agentsTool page
DVAADeliberately vulnerable AI agent for security trainingTool page
ai-trustTrust verification for AI packagesnpm i -g ai-trust
aicomplyPII and credential checks for agent I/OTool page

Install commands are shown only for packages on npm or PyPI as of 2026-08-26; other tools link to their page.