Tutorial: Release Gate with Replay Evidence

Run a production release gate that combines health checks, policy sanity, and replay proof.

Before you start

1. You have a staging environment with near-production data shape.
2. You can run gate scripts and API checks.
3. You have at least one known-good replay run ID.

What you will finish with

A release evidence packet that combines health output, gate output, policy sanity, and replay proof.

Tip - Copy and run Use code block copy buttons directly. If variables are missing, initialize them with One-click Environment Template.

Input

Input fields

Field	Required	Used in steps	Example
BASE_URL	Yes	1, 2, 3	http://localhost:3001
AIONIS_API_KEY	Yes	2, 3	aionis_live_xxx
tenant_id	Yes	2, 3	default
scope	Yes	1, 2, 3	support
known_good_run_id	Yes	3	3d1868e2-e6d3-4f69-952e-61f53ef2ef30

Output fields to persist

Field	Source step	Why keep it
Health response	1	Runtime readiness proof
Core gate summary	1	Release approval gate evidence
applied policy result	2	Pre-release decision consistency
Replay timeline[]	3	Workflow stability proof
run_id, decision_id, commit_uri	3	Compliance-grade provenance chain

Steps

Step 1: Run baseline health and gate checks

curl -sS "$BASE_URL/health" | jq
npm run -s gate:core:prod -- --base-url "$BASE_URL" --scope support

Step 2: Run policy sanity probe

TypeScript

const res = await fetch(`${process.env.BASE_URL}/v1/memory/rules/evaluate`, {
  method: 'POST',
  headers: {
    'content-type': 'application/json',
    'x-api-key': process.env.AIONIS_API_KEY!
  },
  body: JSON.stringify({
    tenant_id: 'default',
    scope: 'support',
    context: { intent: 'billing_support' },
    include_shadow: true
  })
})

console.log(await res.json())

Python

import os
import requests

policy = requests.post(
    f"{os.environ['BASE_URL']}/v1/memory/rules/evaluate",
    headers={"content-type": "application/json", "X-Api-Key": os.environ["AIONIS_API_KEY"]},
    json={
        "tenant_id": "default",
        "scope": "support",
        "context": {"intent": "billing_support"},
        "include_shadow": True,
    },
    timeout=20,
)

print(policy.json())

cURL

curl -sS "$BASE_URL/v1/memory/rules/evaluate" \
  -H "X-Api-Key: $AIONIS_API_KEY" \
  -H 'content-type: application/json' \
  -d '{
    "tenant_id":"default",
    "scope":"support",
    "context":{"intent":"billing_support"},
    "include_shadow":true
  }' | jq '{considered,matched,applied}'

Step 3: Validate one replay chain

curl -sS "$BASE_URL/v1/memory/replay/runs/get" \
  -H "X-Api-Key: $AIONIS_API_KEY" \
  -H 'content-type: application/json' \
  -d '{
    "tenant_id":"default",
    "scope":"support",
    "run_id":"<known_good_run_uuid>",
    "include_steps":true,
    "include_artifacts":true
  }' | jq

Pass criteria:

1. required steps are success
2. no unknown or failed status on gated path
3. provenance identifiers are present (run_id, decision_id, commit_uri)

Step 4: Attach release evidence

Include these artifacts in change approval:

1. health output
2. core gate summary
3. policy sanity output
4. replay timeline with IDs

Expected response sample

{
  "health": { "ok": true },
  "policy_probe": {
    "status": "ok",
    "considered": 5,
    "matched": 2
  },
  "replay_probe": {
    "run_id": "3d1868e2-e6d3-4f69-952e-61f53ef2ef30",
    "required_steps_success": true
  }
}

Common failure and fix

Failure:

gate:core:prod failed: policy check mismatch on support scope

Fix:

1. Re-run rules/evaluate with the exact gate context payload.
2. Check for recently promoted rules in wrong state.
3. Re-validate in staging before go-live decision.

Success criteria

1. Health endpoint and core gate both pass.
2. Policy probe output is stable for the release context.
3. Replay probe confirms required steps as successful.
4. Release packet includes run-level provenance IDs and artifacts.

Rollback trigger guidance

Rollback if any of these happen after release:

1. core gate fails on hotfix run
2. policy decision diverges on critical path
3. replay shows failed or unknown status on required steps