ethz-spylab/agentdojo

A Dynamic Environment to Evaluate Attacks and Defenses for LLM Agents.

69
/ 100
Established

AgentDojo helps developers and security engineers test how well their LLM agents can resist prompt injection attacks and how effective their defenses are. It takes in various attack scenarios and defense strategies, and outputs a benchmark of the agent's resilience. This is for anyone building or securing applications powered by large language models.

471 stars. Available on PyPI.

Use this if you are developing LLM agents and need to systematically evaluate their security against adversarial prompts or compare different defense mechanisms.

Not ideal if you are an end-user looking for a pre-built solution to protect your LLM application, rather than a development and evaluation tool.

LLM-agent-development application-security prompt-engineering security-benchmarking AI-safety
Maintenance 10 / 25
Adoption 10 / 25
Maturity 25 / 25
Community 24 / 25

How are scores calculated?

Stars

471

Forks

118

Language

Python

License

MIT

Last pushed

Mar 12, 2026

Commits (30d)

0

Dependencies

14

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/ethz-spylab/agentdojo"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.