Skip to main content
1

Get your API key

Sign up and create an API key from the Tuteliq Dashboard. Each key is scoped to an environment (production, staging, or development) which determines your rate limits and tier.
Store your API key securely. It is displayed only once at creation time and is hashed server-side.
2

Install the SDK

3

Make your first call

Use detectUnsafe to analyze a piece of text for harmful content.
Expected response:
/safety/unsafe is a starting point, not a complete safety layer. It scores the KOSA harm categories: self-harm, eating disorders, substance use, depression and anxiety indicators, compulsive use, sexual content, violence and hate speech.It does not score everything a child-safety layer needs. Peer harassment, exposure or “fisha” accounts and image-based abuse of a named person are scored by /safety/bullying; grooming trajectories by /safety/grooming; extremist recruitment by /safety/radicalisation.It also returns unsafe: false when the sender is describing a harm rather than being subject to it: survivor testimony, a news article, a lesson, or a child relaying something they saw elsewhere. That is deliberate, and it keeps protective speech out of your review queue. It does mean a message another detector would flag can come back clear here.So unsafe: false means “no KOSA-category signal on this endpoint”, not “this content is safe.” Pair this endpoint with /safety/bullying, or call /analyse/multi and let one request cover several detectors.
Building a pipeline, not a UI? The response above is designed to be read by a person — the full example includes fields like rationale meant for a moderator. If your code branches on risk_score / severity / recommended_action without rendering anything for a human, pass options: { verdict_only: true } (or verdictOnly: true in the SDKs) on any detection call. See Fast mode.
4

Try fraud detection

Use the fraud detection endpoints to analyze content for social engineering, scams, and more.
5

Multi-endpoint analysis

Analyze content against multiple detectors in a single API call using /analyse/multi.
Testing without credits? Create a sandbox API key in your Dashboard (Settings > API Keys > Environment: Sandbox). Sandbox keys run real analysis with zero credit consumption — 50 calls/day for free.
What’s next? Explore the full API Reference or dive into one of our SDKs: Node.js, Python, Swift, Kotlin, Flutter, React Native, .NET, Unity, CLI, or MCP Server.

Next steps

Authentication

Learn about API key environments and authentication methods.

Fast mode

The recommended default for programmatic callers — structured fields without the human-facing formatting.

Credits

Understand per-endpoint credit costs and usage tracking.

Error Handling

Handle errors gracefully with retry strategies.

KOSA Coverage

Full coverage of all nine KOSA harm categories.