What is DKnown AI?
DKnownAI Guard is a specialized AI agent security API designed to protect agentic AI systems from manipulation—not just filter bad content. While traditional moderation tools focus on whether text is offensive or harmful, DKnownAI Guard asks a more critical question: Is someone trying to trick your AI into doing something dangerous? This is essential for AI agents that can execute code, access databases, or manage real-world systems.
Built specifically for the agent era, DKnownAI Guard helps developers block prompt injection attacks, jailbreak attempts, and system hijacking—without breaking legitimate workflows like coding assistance or file management. It gives you precise control by classifying every request into one of four clear risk levels, so you know exactly how to respond.
What are the features of DKnown AI?
- Prompt Injection Detection: Instantly identifies and blocks attempts to override your agent’s instructions using deceptive prompts.
- Four-Level Risk Classification: Categorizes inputs as AGENT_HACK, SYS_FLAG, CONTENT_FLAG, or SAFE—so you get actionable signals, not just alerts.
- Intent-Driven Analysis: Goes beyond keyword filtering to understand the user’s true intent, reducing false positives on legitimate commands.
- Context-Aware Detection: Optionally uses conversation history to improve accuracy while keeping latency low.
- Agentic Workflow Support: Designed for AI agents like Claude Code, Manus, and OpenClaw that interact with real systems (e.g., running scripts or modifying configs).
- Multilingual Protection: Supports over 100 languages out of the box.
- Developer-Controlled Responses: Only AGENT_HACK requests are auto-blocked; you decide how to handle system or content risks.
What are the use cases of DKnown AI?
- Securing an AI coding assistant that executes terminal commands to prevent malicious script injection.
- Protecting a customer support agent with database access from being tricked into exposing user credentials.
- Safeguarding an internal operations bot that manages cloud infrastructure from unauthorized config changes.
- Filtering user inputs in a public-facing AI app to separate harmful manipulation from routine questions.
- Enabling safe file-editing capabilities in an AI agent without allowing deletion of critical system files.
- Complying with enterprise security policies while maintaining full functionality for authorized admin tasks.
How to use DKnown AI?
- Sign up for a free account at DKnownAI Guard and get your API key.
- Integrate the REST API into your AI agent’s input pipeline before processing user requests.
- Send each user message (with optional conversation context) to the DKnownAI Guard endpoint.
- Check the returned classification: AGENT_HACK = block immediately; others = apply your custom logic.
- Use the SYS_FLAG and CONTENT_FLAG signals to trigger human review, logging, or approval workflows.
- Monitor usage via the dashboard and scale to the Pro plan when you exceed 1,000 monthly calls.
