REDAX / INDEX

Documentation

Understand Redax, run it locally, integrate one safe HTTP boundary, and decide what your deployment must provide for production.

WHYKeep raw customer text outside the model boundary.

WHATA local API that detects, replaces, and records safely.

HOWRun it, send text, pass only returned text onward.

New to Redax?

Start with the local Docker deployment, then send a synthetic request through the API. The product is a boundary in front of your existing model workflow — not a replacement for your application or provider.

Open the deployment guide →

Choose your next step

Integrate the APISee the request and response contract, then put redaction immediately before your model call. Copy an exampleRun a local request, call the API from Python, and configure Redis, telemetry, and production secrets. Understand the boundaryTrace detection, policy, replacement, audit, Redis state, worker jobs, and telemetry. Deploy locallyRun the complete Compose stack with Redis, worker, metrics, logs, and traces — no cloud account required. Plan productionWalk through hosting, Redis HA, secrets, telemetry, DNS/TLS, SLOs, failure drills, and stop conditions. Configure the runtimeFind every REDAX_* setting with its type, default, example, production requirement, and restart behavior.

What Redax does

Redax detects configured entities, applies a versioned policy, and returns transformed text. Structured regex detection is the stable lightweight path; the pinned local GLiNER2 path adds contextual entities when its model is available. The browser demo is intentionally illustrative and does not claim server-side parity.

Read the boundary before sending real data.

Detection is configured coverage, not proof of absence. Review the threat model, data flow, failure modes, policy pass-through behavior, provider logging, infrastructure access, and retention before production use.

Read the security boundary →