Red-team strategy

Context Compliance (fabricated assistant turn)

context-compliance

Plants a fake prior turn where the assistant already answered a benign version of the request and offered to continue; the model stays consistent with the planted history and complies with the harmful follow-up.

YAML config

evalguard.config.yaml

redteam:
  strategies:
    - context-compliance

CLI

bash

# "context-compliance" 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.