Skip to content

Latest commit

 

History

History
17 lines (12 loc) · 1.55 KB

File metadata and controls

17 lines (12 loc) · 1.55 KB

Stack Skills

5 cognitive firewalls that defend against specific reasoning failures.

Skills

  • adversarial-review -- Devil's Advocate stress-testing. Steel-mans, then attacks via 3 vectors (logical soundness, edge cases, structural integrity). Use before merging significant PRs or committing to architecture decisions.
  • creativity-sampler -- Probability-weighted option generator. Produces 5 alternatives across a typicality spectrum (conventional to wild card) and exposes hidden assumptions. Use at decision points when you catch yourself defaulting to the obvious answer.
  • cross-verified-research -- Source-traced research with anti-hallucination safeguards. Every claim must be cross-verified by 2+ independent sources with S/A/B/C tier grading. Use when accuracy matters more than speed.
  • pre-mortem -- Prospective failure analysis. Assumes complete failure, works backward through 5 categories (technical, organizational, external, temporal, assumption) to surface risks with leading indicators and circuit breakers. Use before committing to plans or irreversible decisions.
  • reasoning-tracer -- Makes reasoning chains auditable. Forces assumption inventories, decision branching, confidence decomposition, and weakest-link identification. Use when the user needs to see WHY a conclusion was reached.

Recommended Chains

  • Tech Decision: creativity-sampler -> cross-verified-research -> adversarial-review
  • Architecture Review: reasoning-tracer -> adversarial-review
  • Project Kickoff: pre-mortem -> creativity-sampler (for mitigations)