DKnownAI Guard
Security API that guards agent based AIDetect prompt injection and agent hijacking early
Highlights
What you can do with it
- Checking a Claude Code based AI agent for prompt injection attempts
- Watching a Manus AI agent that manages databases for risky system operations
- Screening an OpenClaw AI agent for content compliance issues
How to use it
- 1 Visit the DKnownAI Guard site, choose a package, and register an account.
- 2 Integrate the API into your own AI agent system using the documentation.
- 3 Send incoming requests to DKnownAI Guard for analysis.
- 4 Act on the result, whether SAFE, CONTENT FLAG, SYS FLAG, or AGENT HACK.
- 5 Review logs and statistics regularly to track security status.
Who it is for
Pros
- Purpose built for agent manipulation rather than generic filtering
- Clear four level results tell you what action to take for each request
- The free package includes 1000 API calls per month
Limitations
- The free package is capped at 1000 API calls per month
- Detection depends on the request context, so classification can vary
- Package pricing and free limits may change over time
What DKnownAI Guard is
DKnownAI Guard is a security API designed specifically for agent based AI. It is not traditional content review. Its main job is to detect the manipulation behavior of AI agents, such as prompt injection, jailbreaking, and agent hijacking, so that agents that execute code, modify databases, or manage server configurations cannot be taken over by an attacker. The service uses an intention driven approach instead of simple word filtering, and it classifies each request into one of four categories: AGENT HACK, SYS FLAG, CONTENT FLAG, or SAFE. You can choose to include conversation context for detection, which improves accuracy while keeping latency low, and the product fully supports 100 languages. It also protects agent autonomy, so legal operations still run normally while attacks are blocked. The product is positioned to provide security solutions for AI agent developers and production teams.
What you can do with it
- Check agent requests for prompt injection and jailbreaking attempts
- Flag risky system operations before an agent changes a database
- Screen agent output for content compliance issues
- Use a clear decision tree to choose the right action for each request
Who it is for
- AI agent developers building agents that take real world actions
- Production teams responsible for agent stability in live environments
- Enterprises that use AI agents for code execution and database work
What to watch out for
The free package is limited to 1000 API calls per month, so busy agents will need the professional package. Classification quality depends on the request and the context you send, so borderline cases may be labeled differently between attempts. This is a security layer, not a guarantee, so keep your own logging and review process in place, and note that package pricing and free limits may change over time.
FAQ
What does DKnownAI Guard detect?
It detects manipulation of AI agents, including prompt injection, jailbreaking, and agent hijacking.
What results does it return?
Each request is classified as SAFE, CONTENT FLAG, SYS FLAG, or AGENT HACK.
Is there a free option?
Yes. The free package costs $0 per month and includes 1000 API calls.
Why it is free
What the official page says： 免费免费套餐
Free limits, trial length and features are decided by the official site and can change at any time.
Product preview
Screenshot archived from the listing page, for preview only.
