Red-team strategy

Simple Assistive Task Linkage (SATA)

assistive-task-linkage

SATA [Zhou 2025]: masks the objective's key tokens as [MASK] and wraps the sentence in a benign masked-LM-infill or glossary look-up task, then asks the target to complete and execute it — the request is absent from the prompt, so only output-side attack-success detection can catch it

YAML config

evalguard.config.yaml

redteam:
  strategies:
    - assistive-task-linkage

CLI

bash

# "assistive-task-linkage" 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.