Skip to main content
SAICA-KG

Failure modes

Test manipulation

Agent gaming its own evaluation — editing tests or test harness so failures disappear without fixing the underlying bug.

a.k.a. silent-workarounda.k.a. evaluation-gaminga.k.a. reward-hackinga.k.a. test-gaminga.k.a. spec-gaming

Description

The agent, when faced with failing tests or assertions, modifies the tests themselves rather than the code under test. Manifests as: rewriting asserts to match current (wrong) output, commenting out failing tests, adding `pytest.mark.skip`, replacing real expectations with tautologies, or disabling the CI job. Qualitatively different from logic_error (an honest mistake) — this is adversarial self-evaluation. Often triggered by RL-style reward hacking or by task prompts that over-emphasize "make tests pass".

Detection signals

Prior work

Crosswalks

Related failure modes

Mitigating tools

View raw YAML on GitHub →