AI Plugins
AI Plugins in DTX enable the system to connect with external tools, prompt libraries, and agent behaviors to extend red teaming capabilities. These plugins offer modularity and composability, helping you simulate real-world attack surfaces and workflows.
What Are AI Plugins?
AI plugins are extensible integrations that allow DTX to:
- Pull and test prompts from public libraries (e.g., LangChain Hub)
- Interface with external LLM agents or platforms
- Dynamically inject logic, context, or transformations into prompt flows
Plugins abstract away backend complexity and enable reusable red teaming strategies across tools and services.
Examples of AI Plugins
| Plugin | Description |
|---|---|
| Langhub | Load prompt templates from LangChain Hub for testing |
| LangGraph | (Upcoming) Load multi-step LLM workflows as agents |
| OpenTools | (Upcoming) Extend agents with tool use (e.g., search, API calls) |
Supported Plugin Types
- Prompt Libraries (e.g., LangChain Hub)
- Tool Extensions (via HTTP/Graph APIs)
- Agent Simulators (planned integrations for agent chains)
DTX aims to provide seamless plug-and-play support for emerging LLM agent platforms and prompt engineering workflows.