Verdicter
Verdicter
Runtime security for AI agents

Your agents act.
Verdicter decides.

A firewall for AI agents. Every tool call is evaluated against your policies before it touches a real system.

ALLOWDENYESCALATEMODIFY
Start for freeRead the docs →
POST /v1/evaluate

Request

{
  "agentId":    "support-bot",
  "actionType": "delete_data",
  "payload": {
    "table":     "users",
    "condition": "id = '123'",
    "bulk":      false
  }
}

Response · 18ms

{
"decision": "escalate",
"riskScore": 90,
"reason": "Role mismatch: support → delete_data",
"evaluationId": "eval_9f2a…"
}
Live evaluation - click a decision tab above

Works with every agent framework

OpenAI
Anthropic
Bedrock
LangChain
CrewAI
Vercel

Why now

Agents are shipping.
Guardrails aren't.

LLM agents call tools, send emails, move money, and touch production. A prompt injection or hallucinated argument is one tool call away from a serious incident. Verdicter sits between your agent and the world.

Tool calls outpace review

Agents make thousands of decisions per minute. Human review doesn't scale.

Prompt injection is real

Adversarial inputs bypass system prompts. You need enforcement at the boundary.

Auditors are asking

SOC 2 and emerging AI regulations require provable controls on agent actions.

Platform

Everything you need to put agents in production.

Enforce

A declarative policy engine evaluates every tool call in under 50ms. Express rules as code or natural language - Verdicter compiles them into deterministic checks.

ALLOW
DENY
ESCALATE
MODIFY

Observe

Full session traces. See every tool call, decision, and policy hit in real time.

Shield

Built-in prompt injection and jailbreak detection on inputs and tool arguments.

Comply

Generate SOC 2 and GDPR-ready reports from your audit log. Every decision is signed and immutable.

SOC 2 GDPR HIPAA-ready ISO 27001

Sandbox

Replay production traffic against new policies before you ship them.

Identity

Track which credential each tool call used. Rotate, scope, and revoke from one place.

How it works

One call.
Before every action.

Drop the SDK in front of any tool your agent calls. Verdicter returns a decision in milliseconds. You decide what to do with it - but the audit trail is already written.

  • 1Install the SDK in 30 seconds
  • 2Wrap your tool calls with evaluate()
  • 3Write policies in code or plain English
  • 4Ship - every action is now governed
agent.ts
const res = await verdicter.evaluate({
  agentId:        "support_bot_prod",
  tool:           "send_email",
  payload:        { to: user.email, subject, body },
  credentialName: "SENDGRID_KEY",
});

if (res.decision === "ALLOW") await sendEmail(payload);

Pricing

Simple, transparent pricing

Hobby

Free

For personal projects and exploring Verdicter.

  • 10,000 evaluations / month
  • npm SDK
  • 3 registered agents
  • 5 active policies
  • 7-day audit log retention
  • Runtime enforcement
  • Shield prompt scanning
  • Sandbox (5 scenarios)
  • Community support
Get started

Starter

$29/month

For solo developers shipping agents to production.

  • 50,000 evaluations / month
  • npm SDK + direct REST API
  • 10 registered agents
  • 20 active policies
  • 30-day audit log retention
  • Runtime enforcement
  • Observe: session tracing
  • Shield: custom injection rules
  • Sandbox (50 scenarios)
  • Identity: credential tracking
  • Email support
Get started

Pro

$79/month

For teams running AI agents in production.

  • 250,000 evaluations / month
  • npm SDK + direct REST API
  • Unlimited agents
  • Unlimited policies
  • 90-day audit log retention
  • Runtime enforcement
  • Observe: session tracing
  • Shield: custom injection rules
  • Comply: SOC 2, GDPR, HIPAA reports
  • Sandbox: unlimited scenarios
  • Identity: credential tracking
  • Team members (shared workspace plan)
  • Webhook notifications
  • Email support
Get started
All prices in USD · Secure payment with
See full pricing & feature comparison

FAQ

Common questions

What counts as an evaluation?

Every call to verdicter.evaluate() - or a direct POST to /v1/evaluate - counts as one evaluation, regardless of the decision (ALLOW, DENY, MODIFY, or ESCALATE).

Does Shield scanning count toward my quota?

No. Shield scans for prompt injection are tracked separately and never deducted from your evaluation quota.

What's the difference between the SDK and the REST API?

The npm SDK wraps the REST API with TypeScript types, retries, and adapters for LangChain and Vercel AI SDK. The raw REST API is available on Pro for teams using Python, Go, or any other language.

What happens when I hit my monthly limit?

Additional evaluations are blocked until your quota resets on the 1st of the month. You can upgrade at any time to restore access immediately.

Is my agent payload data stored securely?

Payload data lives in your own Supabase project - only you control it. Verdicter never has independent access beyond what your own deployment exposes.

Can I upgrade or downgrade at any time?

Yes. Changes take effect at the start of your next billing cycle.

Start in five minutes.

Install the SDK, wrap one tool call, and you'll have your first audit trail before your coffee gets cold.

Start for freeTalk to engineering