01 · Before deployment
Agent Verify®
Scan agents, tools, and MCP servers for security risks before they ship.
The AI agent security platform
Secure what your AI agents execute.
One control layer across the agent lifecycle: verify agents before they deploy, authorize every action at execution, and prove what each action delivered.
02 · At execution
A2SPA™
Authorize every protected action. No signature, no execution.
03 · After execution
A2EA™
Measure what every agent delivers: success rate, time saved, and dollar value.
Available through
Built on
Google Cloud
SAM registeredCAGE 9PH50
Patents pendingA2SPA & A2EA
Built on
Google Cloud
SAM registeredCAGE 9PH50
Patents pendingA2SPA & A2EA
Why now
AI agents now take real actions. Logins don't control them.
Agents act, not just answer
They move money, change records, call tools and MCP servers, and trigger workflows, often with no person reviewing each step.
Identity isn't authorization
SSO and API keys prove who an agent is. They don't decide whether this specific action, with this exact payload, should run.
Every action needs proof
Security reviews, auditors, and budget owners want evidence of what each agent did, who approved it, and what it was worth.
AI Modularity covers the whole lifecycle: Agent Verify before deployment, A2SPA at execution, and A2EA after. Use it alongside Okta or your existing identity provider.
Products
Three products. One platform.
Verify agents before they deploy, authorize every protected action, and prove the value they deliver.
Agent Verify® 2.0
Find exposed secrets, risky permissions, unsafe MCP setups, and prompt-injection holes in your AI agents before they ship.
- 124 tested checks on agent code, tools, and MCP config
- Shows the exact line and the fix; Pro re-checks your changes with AI
- Checks every pull request on GitHub, GitLab, and Bitbucket
- Works in VS Code, Claude Code, Cursor, and CI
- Mapped to OWASP LLM Top 10, NIST AI RMF, and SOC 2
A2SPA™ 2.4.2
Enforce authorization before protected actions execute.
- Every protected action is cryptographically signed
- Unsigned, altered, or replayed actions are blocked
- A verifiable receipt for every decision
A2EA™ 2.0
Measure and prove the value of AI agent activity.
- Tracks every completed agent task
- Time saved, success rate, and dollar value
- Reporting ready for budget and audit reviews
Demo Center · 24/7
See it work.
Three products, one flywheel. Watch each demo, then run a live scan or open the live dashboard. No sign-up needed.
Agent Verify® · Verify
See how agents, tools, and skills are scanned for security risks before they ship.
A2SPA™ · Authorize
Watch execution-time authorization stop any agent action that wasn't signed and approved.
A2EA™ · Prove
Watch agent attribution, performance measurement, and ROI reporting in action.
Want a guided demo for your team? Request a demo →
Buy Now
Start today. Scale when you're ready.
Self-service plans to get started, and enterprise licensing through sales or procurement.
Agent Verify
Scan and fix agent risk
Free
$0 / month
10 scans a month, all 124 checks, and pull request checks.
Pro · 14-day free trial
$19.99 / month
100 scans, full fixes, corrected code, and PDF reports.
Team
$25 / seat / month
Pro for your whole team, from 2 seats.
A2SPA
Authorize every protected action
Self-service
Get started online
Create an account, add your agents, and connect with the integration pack.
Enterprise license
From $250,000 / year
Production deployment, security review, and procurement support.
A2EA
Prove agent value
Free test
$0 first 100 events
Connect one agent function and see the value dashboard.
Usage
Pay as you grow
Keep sending events after the free test. Enterprise plans available.
Government and enterprise procurement
Purchase through Carahsoft, or contact sales for a quote and security review.
SAM registered
UEI UCJJGZLA1PQ4 · CAGE 9PH50
Carahsoft
Public-sector distribution
Google Cloud
Production infrastructure
Patents pending
A2SPA and A2EA
FAQ
Questions buyers ask.
How is AI Modularity different from Okta or other identity tools?
Okta and similar tools manage human and workforce identity: who someone is and what they can sign in to. AI Modularity controls what an AI agent is allowed to execute, one action at a time. The two work together: keep your identity provider, and add A2SPA to authorize each protected agent action.
How is A2SPA different from prompt guardrails?
Guardrails filter what goes into and comes out of a model. A2SPA authorizes the action itself, at the moment it would execute. Each action is signed, and the signature covers the sender, target, timestamp, nonce, and exact payload. A2SPA checks the signature, your policy, and freshness, blocks unsigned, altered, or replayed actions, and issues a receipt for every decision. A target-side guard fails closed.
Read the A2SPA docs →Which agent frameworks and tools does Agent Verify support?
Agent Verify includes checks for 11 agent frameworks: LangChain, LangGraph, CrewAI, AutoGen, LlamaIndex, OpenAI Agents SDK, Claude Agent SDK, Vercel AI SDK, Semantic Kernel, smolagents, and Google ADK. It also checks MCP server configurations. You can run it from the web dashboard, on pull requests in GitHub, GitLab, and Bitbucket, in VS Code, from Claude Code or Cursor through its MCP server, or from the command line.
See the developer tools →Which compliance frameworks does Agent Verify map to?
Findings are mapped to the OWASP LLM Top 10, NIST AI RMF, and SOC 2, with evidence packs for the EU AI Act and ISO/IEC 42001.
Browse all 124 checks →How do I measure the ROI of my AI agents?
Use A2EA. Wrap an agent function so it sends task events, and the A2EA dashboard shows success rate, time saved, and dollar value for every completed task. The first 100 events are free.
Open the A2EA live demo →Can government agencies and enterprises buy AI Modularity?
Yes. AI Modularity is SAM registered (UEI UCJJGZLA1PQ4, CAGE 9PH50) and available through Carahsoft. Enterprise and public-sector buyers can also request a quote and a security review from our team.
Government contracting details →How do I get started?
Run a free Agent Verify scan on your agent code; the free plan includes 10 scans a month with all 124 checks. You can also create a self-service A2SPA account or start a free A2EA test. For enterprise licensing or a guided demo, contact our team.
Scan your agent free →Have a different question? Ask our team →
Contact sales
Talk to our team.
Enterprise licensing, a guided demo, or procurement questions. Tell us what you need and we'll reply within one business day.
- Enterprise licensing for A2SPA
- Guided product demos
- Security reviews and procurement