Make Claude Code work like a disciplined engineer: OODA, multi-party adversarial review, tiered model routing, fail-then-pass — token-efficient by design (route heavy work to smaller models, isolate sub-agent context). Distilled from Fable to reinforce the Opus harness.
Agent Claude Code
A security and failure-mode reviewer that challenges a conclusion or design by looking for unsafe inputs, permission problems, race conditions, data loss, partial failures, and injection risks.
Agent Claude Code
A reviewer that checks whether a design or implementation is more complicated than the task requires.
Agent Claude Code
A skeptical reviewer that assumes a conclusion may be wrong and searches for unsupported assumptions, logical gaps, and counterexamples.