Spaces:
Running
Running
| title: AgentInjectionBench | |
| emoji: 🔬 | |
| colorFrom: red | |
| colorTo: purple | |
| sdk: gradio | |
| sdk_version: "5.29.0" | |
| python_version: "3.11" | |
| app_file: app.py | |
| pinned: false | |
| # AgentInjectionBench | |
| **The first benchmark for prompt injection attacks in agentic tool-use pipelines.** | |
| **v0.1 — 120 hand-crafted seed samples** across 6 attack categories. The target is **2500+ samples** and we welcome community PRs to help get there. | |
| 👉 **[Contribute on GitHub](https://github.com/ppradyoth/AgentInjectionBench)** — add seed templates, run the generation pipeline, or add a new LLM provider (Gemini Flash, Ollama, HF Inference API most wanted). | |
| - [Dataset](https://huggingface.co/datasets/ppradyoth/AgentInjectionBench) | |
| - [GitHub](https://github.com/ppradyoth/AgentInjectionBench) | |