---
title: FAQ
description: "Frequently asked questions about Syntext."
---

# FAQ

## General

### How is Syntext different from Mintlify or ReadMe?

Three things: **code annotations** (API reference generated from doc comments in your actual source, with drift detection when code and docs diverge), an **AI assistant** that reports what it *couldn't* answer as a ranked gap list, and **content health analytics** that treat docs as a product with metrics. Content is standard MDX in git — no proprietary editor lock-in.

### Do I have to change my code to use Syntext?

No. Existing JSDoc, docstrings, Go doc comments, Javadoc, PHPDoc, and C# XML docs are parsed with zero configuration. `@stx` directives are optional enhancements — see [Annotations](/concepts/annotations).

### What languages are supported for annotation parsing?

TypeScript/JavaScript, Python, Go, Rust, Java/Kotlin, PHP, and C#. [Full matrix](/concepts/annotations).

### Can I self-host?

The doc *site output* is static and portable, and a [self-hosted deployment guide](/deployment/self-hosted) covers running the renderer yourself. The managed platform (builds, AI, analytics, collaboration) is hosted by Syntext.

## Content

### Is my content locked in?

No. Your docs are plain MDX files in your own repository. Leave anytime — your content already lives with you.

### Can I keep some pages private?

Yes — three mechanisms: [audience gating](/guides/audience-gating) for tiered access (e.g. partners), path-based [access rules](/platform/collaboration), and IP allowlists.

### Does Syntext support versioned docs?

Use git branches: each branch can build previews, and production tracks your configured branch. Dedicated first-class version switching is on the roadmap.

### Can non-developers edit docs?

Yes — the [web editor](/platform/web-editor) gives writers a browser-based editor with live preview, AI assistance, linting, and review workflows. No git required.

## AI

### Is my content used to train models?

No. Content is embedded for retrieval within your project's isolated namespace only. It is never used for model training and never shared across tenants.

### What happens when the AI doesn't know an answer?

It says so rather than inventing one, and the question is logged as a **documentation gap** in [analytics](/platform/analytics) so you can write the missing page.

### Can I put the AI assistant on my marketing site or in my app?

Yes — one script tag. See the [widget recipe](/recipes/ai-widget-anywhere).

## Billing & Limits

### What counts against my plan?

Builds (including PR previews) and AI queries, metered monthly per organization. Current usage is under **Organization → Billing**; limit behavior is documented in [Rate Limits](/api-reference/rate-limits).

### What happens if I hit my AI query limit?

The assistant shows a friendly limit message to visitors, and the API returns `403 USAGE_LIMIT_EXCEEDED`. Your doc site, search, and everything else keep working normally.

## Security

### How are API keys stored?

Only a SHA-256 hash is stored — the full key is shown once at creation. Keys support scopes (`full`, `build`, `read`), expiry, [rotation](/api-reference/api-keys/rotate), and revocation.

### Are webhooks verified?

Yes — inbound GitHub webhooks are verified with HMAC-SHA256 timing-safe signature checks. Unsigned or mis-signed deliveries are rejected.