Add a decision layer to AI agent systems

See how Neura evaluates AI actions, returns a Decision Receipt, and fits into your agent stack

Try PlaygroundRun a local check
Diagram showing an agent action entering Neura Relay authority checks and returning a Decision Receipt

Playground

Run a representative AI action, inspect the decision and receipt, and see what your runtime receives

Action

Enter the action an agent wants to take

Request
await relay.resolve({
  action_card: {
    version: "0.1",
    proposedAction: {
      summary: "<proposed action>"
    }
  }
})

Review

Runs after you submit the proposed action

Decision

Unlocks when Relay returns the decision

Run a local check

See how Relay reviews a proposed action before execution

Use this command to clone the repo, run the local check, and inspect the Decision Receipt

Copy and run locally
git clone https://github.com/neurarelay/relay-action-card.git
cd relay-action-card
npm ci
npm run first-proof -- --dry-run --source=builders --campaign=package_reality_first_proof --surface=builders_first_proof --json
About the local check
No account

No account or token needed

No private data

Uses sample data only

Relay reviews

Returns a Decision Receipt

Your runtime decides

Chooses what happens next

Use this after the local check to create a real Decision Receipt reference for your evaluation

Create a live receipt (optional)
npm run first-proof -- --source=builders --campaign=package_reality_first_proof --surface=builders_first_proof --json
Open the Relay Action Card quick start

Choose your integration path

Connect Relay to the runtime you already use

Need protected access or hosted evaluation? Open Developer Workspace

Questions builders ask

The practical answers before you connect a first action

Ready to evaluate a real action in your stack?

Evaluate the workflow your team owns and define the integration and review path

Keep credentials and live data out of the first evaluation

Evaluate Neura