Back to browse
GitHub Repository

Behavioral governance framework for AI agents. Named anti-patterns, incident-driven rules, and failure mode engineering.

11 stars

Named failure modes that stop AI agents from cutting corners

by travisdrake·Mar 24, 2026·1 point·0 comments

AI Analysis

MidBig BrainNiche Gem

Named failure modes for AI agents, but it's markdown files—not an actual tool or implementation.

Strengths
  • 11 named anti-patterns traced to specific incidents give concrete behaviors to match against.
  • Extends Karpathy's context engineering definition into the behavioral governance layer.
  • Six standalone rules ready to copy directly into agent configurations.
Weaknesses
  • Pure documentation repo—no software implementation, automation, or enforcement mechanism.
  • Conceptual framework without tooling means adoption depends entirely on manual discipline.
Category
Target Audience

Developers building or configuring AI agents

Similar To

AI Safety Frameworks · Constitutional AI · RLHF Guidelines

Similar Projects

AI/ML●●●Banger

LoopGain – Cut agent API spend by measuring when loops stop improving

1921 electrical engineering oscillator math stops AI loops at convergence.

Big BrainWizardry
fitz2882
213d ago