ravenoak-llm-skills

Use when stress-testing a conclusion, auditing a complex claim, mapping a multi-actor system, or spotting motivated reasoning in a single recursive pass — when rigor matters more than speed. Skip for time-pressured decisions, mechanical tasks, lookups, or when the user wants a recommendation rather than a tour of considerations. For a decision with genuine multi-stakeholder stakes spanning several teams, or organization-scale, hard-to-reverse consequences, use reasoning-swarm instead — its independent-parallel-agent dispatch is worth the extra cost there.

Use when a decision or proposal has genuine multi-stakeholder stakes, spans multiple teams, or carries organization-scale or hard-to-reverse consequences — build-vs-buy calls, architecture decisions affecting several teams, policy with systemic impact, or when the user explicitly wants independent parallel takes stress-tested against each other rather than one recursive pass. Skip for single-claim analysis or solo stress-testing (use reasoning-framework instead), time-pressured decisions, mechanical tasks, or when a recommendation is wanted rather than a tour of considerations.