- Pricing
- Freemium
- Free, Pro
- Added
- Apr 21 2026
About DKnownAI Guard
DKnownAI Guard is a security API built for agentic AI, used by developers to detect manipulation of AI agents that execute code, access files, call tools and interact with real systems. It is designed for agents such as Claude Code, Manus and OpenClaw, protecting agent autonomy without blocking legitimate operations.
A developer sends a request's text to one REST API endpoint with an API key, and DKnownAI Guard analyzes intent rather than just filtering words, returning one of four risk classifications: AGENT_HACK, SYS_FLAG, CONTENT_FLAG or SAFE. An optional session_id lets the API evaluate a message together with earlier turns in the same conversation to catch multi-turn escalation.
DKnownAI Guard separates manipulation attempts such as prompt injection, jailbreak and system prompt extraction (AGENT_HACK) from high-risk system operations like deleting a database (SYS_FLAG) and from compliance-sensitive content (CONTENT_FLAG), leaving routine requests marked SAFE to process normally. It supports more than 100 languages and includes context-aware detection for multi-turn conversations.
DKnownAI Guard offers a Free plan at $0 a month with 1,000 API calls, and a Pro plan at $1.99 a month (first month $0.99) with 5,000 API calls and priority technical support, marked as coming soon. It is accessed over a REST API endpoint and is operated by Shenzhen Dknown Intelligence Technology Co., Ltd., with its AI models registered with China's Cyberspace Administration.
Use cases
- Blocking an agent hijacking attempt impersonating an administrator
- Flagging a risky system operation like deleting a production database for developer review
- Detecting a multi-turn social engineering attack that escalates to credential theft
- Protecting developer tools, copilots and automation agents that trigger real actions
Key features
- Detects prompt injection and jailbreak attempts
- Four-level intent-driven risk classification
- Context-aware detection using conversation history
- Separates operational risk from content risk
- Supports more than 100 languages
- REST API with Bearer token authentication
- session_id tracks multi-turn conversation context
- Sub-200ms response latency
- Proprietary safety classification model trained on multilingual content
- Native integration with the OpenClaw agent framework
Pros and cons
Pros
Cons
Pricing
- 1,000 API calls / month
- All four safety classifications
- Multilingual support
- REST API
- Email support
- First month $0.99
- 5,000 API calls / month
- All Free features included
- Priority technical support
- Built for production teams
Reviews
Frequently asked questions
What is DKnownAI Guard?
How is DKnownAI Guard different from content moderation?
Does DKnownAI Guard detect prompt injection and jailbreak attempts?
What do AGENT_HACK, SYS_FLAG, CONTENT_FLAG, and SAFE mean?
When should I use session_id in requests?
Alternatives to DKnownAI Guard
More in AI Developer Tools
Adaptive Engine
Adaptive Engine evaluates, tunes, and serves LLMs for enterprise AI, using reinforcement learning to optimize models and production feedback.
Prompt Octopus
Prompt Octopus compares model responses side-by-side, with a VSCode extension and API key options for use in your codebase.
Antigma
Antigma provides self-contained, self-organizing agents for developers, with Ante as an autonomous coding agent that runs locally or cloud-backed, in a terminal and via various interfaces.
Lingo.dev
Lingo.dev is a localization engineering platform for product teams to measure translation quality, choose translation providers, and manage localization…
Zipy
Zipy provides AI-powered session replays, error tracking, performance monitoring, and product analytics for developers and product teams.
Hyper AI
Hyper AI runs live, voice-driven product demos and in-app onboarding for SaaS companies, with an AI agent that controls the real product on screen.