Red-team strategy

Dynamic Jailbreak

dynamic-jailbreak

Generates novel jailbreak prompts by combining persona, scenario, and instruction override techniques

YAML config

evalguard.config.yaml

redteam:
  strategies:
    - dynamic-jailbreak

CLI

bash

# "dynamic-jailbreak" is layered onto attacks during red-team scans.
# List every available encoding/obfuscation strategy:
evalguard list strategies

Combine with attack plugins to generate adversarial payloads. The strategy wraps each base payload before it's sent to your target, testing whether the model decodes / unwraps the transform and follows the embedded instruction.

Threat-model fit

Mutation-based generation of jailbreak variants (GCG/GPT-fuzzer-style). Useful when you've patched a specific attack and want to test variants. Treats the model as a black-box function and mutates toward higher harm scores.

Pairs well with

tree-search best-of-n composite-jailbreaks

Hand-written · 2026-05-21