Control what AI systems are allowed to do before actions execute

Evaluate, permit, deny, escalate, or silence machine actions before they change production systems - with receipts you can verify offline.

AI wants to act. TrigGuard decides whether execution is authorized.

Execution posture

Fail-closed by default

Fail-closed execution model.

Cryptographic receipts

Receipts verifiable offline.

Built for critical systems

Auditable execution posture.

Authorize before irreversible action

Automation proposes. TrigGuard decides. Protected surfaces execute only on PERMIT.

  1. Intent Agent or automation proposes a protected action
  2. Authorize Gateway evaluates policy · PERMIT · DENY · ESCALATE · SILENCE
  3. Execute Only PERMIT unlocks the irreversible side effect

EXECUTION SURFACES

  • Deployments
  • Transfers
  • Exports
  • External APIs
  • Infrastructure
  • Robotics

Explore Architecture · Products · Decision model

Production deployment posture

<5ms

Decision latency

Kernel hot path · p99

<15ms

End-to-end authorization

Finite-state eval · p99

  • Fail closed
  • Receipts
  • Deterministic authorization
  • Offline verification

Built for environments where execution has consequences.

  • Every decision is provable.
  • Every policy is versioned.
  • Every execution is governed.

Your first authorized decision
In under 2 minutes

Quick verification uses the offline demo. Live authorization uses the same SDK with TRIGGUARD_API_KEY. Full walkthrough: Quickstart.

  1. Install
  2. Authorize
  3. Receipt
  4. Verify
Your Agent

Install

Add the SDK, then run the offline demo for a first verified receipt.

npm install trigguard
npx trigguard demo
pip install trigguard
# Same first verified receipt (shared CLI):
npx trigguard demo
TrigGuard Authorize

Authorize

For live authorization, set TRIGGUARD_API_KEY, then request a decision for a protected surface.

# export TRIGGUARD_API_KEY=tg_live_…
import { authorize } from "trigguard";

const result = await authorize({
  surface: "deploy.release",
  actor: "my-app",
});
# export TRIGGUARD_API_KEY=tg_live_…
from trigguard import authorize

result = authorize(
    surface="deploy.release",
    actor="my-app",
)
Receipt PERMIT · Signed

Receipt

A live PERMIT (or the offline demo) returns a receipt.

[ DECISION ]
{
  "decision": "PERMIT",
  "executionId": "…",
  "receipt": { "receiptHash": "…", "authoritySignature": "…" }
}
Verify Independent

Deploy TrigGuard as a gateway, sidecar, or policy enforcement integration based on runtime constraints and rollout posture.

Execution authorization in your stack

  • Zero Standing Access
  • Cryptographic Receipts
  • Open & Verifiable
  • Cloud & On-Prem Ready
  • SDKs & APIs

These systems execute. TrigGuard authorizes.

All integrations · Node/Express · MCP

Select an action, view the decision

Simulated locally. Each selection returns a decision and receipt fields - no backend required.

Select an action to view the authorization decision.

Signed execution receipts

Signed on every decision path. Verify offline without a vendor log pipeline.

Permit Payments

Refund within policy threshold

Execution ID
rcpt_c9d8e7f6a5b43210
Surface
payments.refund
Risk
Medium
Authority
policy:payments-refund@v1
Timestamp
2026-06-03T14:22:08Z

ed25519:1Qp…k9Wz · verified

Deny Infrastructure

Production database deletion blocked

Execution ID
rcpt_7f3a9c2e1b8d4f6a
Surface
infrastructure.delete
Risk
Critical
Authority
policy:prod-no-destructive@v3
Timestamp
2026-06-03T09:41:33Z

ed25519:8Kx…m2Pq · verified

Escalate CI/CD

Production deploy requires approval

Execution ID
rcpt_a1b2c3d4e5f67890
Surface
cicd.deploy
Risk
High
Authority
policy:deploy-governance@v2
Timestamp
2026-06-03T16:05:17Z

ed25519:3Nm…v7Rt · approval required

Start with the console or join the design partner program

Review decisions, receipts, and audit trails in the TrigGuard Console. Developers: install the CLI.

Enterprise pilots: 4-6 weeks, receipt-based evidence, no production replacement required. Procurement center.

Trusted Infrastructure

See partners →