Release Tracking

Release tracking automatically compares your API's performance before and after each deployment, giving you a clear answer to: "Did this deploy make things better or worse?"

Setup

Pass the current version as the release option:

Node.js

app.use(apiforge({
  mode: 'local',
  release: process.env.npm_package_version,
}))

Python

import os
app.add_middleware(
    ApiForgeMiddleware,
    mode="local",
    release=os.environ.get("RELEASE"),
)

Or hardcode it during the deploy process:

Node.js

app.use(apiforge({ mode: 'local', release: 'v1.4.0' }))

Python

app.add_middleware(ApiForgeMiddleware, mode="local", release="v1.4.0")

Automate with environment variables

Most CI/CD systems expose the version or git tag as an environment variable. Inject it at build time so it changes automatically on every deploy.

How it works

1. When the release value changes (new deploy detected), APIForge records a baseline snapshot of the last 24h of metrics per endpoint.
2. After 30 minutes of post-deploy traffic, the engine compares new metrics against the baseline.
3. A PERF or OK insight is generated for each endpoint with enough traffic.

What is compared

Metric	Compared
P50 latency	Before vs after
P90 latency	Before vs after
P99 latency	Before vs after
Error rate (4xx + 5xx)	Before vs after

Example insights generated

✅ OK   v1.4.0 (deployed 3h ago) — No regression detected.
         Latency stable across 12 endpoints. Error rate unchanged.

⚠️ PERF  POST /checkout — P99 +38% after v1.4.0
         Before: 210ms · After: 290ms
         Consider reviewing changes to the checkout handler.

✅ OK   GET /products — P90 improved by 12% after v1.4.0
         Before: 85ms · After: 75ms

Requirements

• The release value must change between deploys (same value = no new baseline)
• At least 100 requests on an endpoint are needed for a statistically valid comparison
• The comparison window is 30 minutes minimum post-deploy