prior-art

What problem does it solve? When designing non-trivial algorithms, protocols, or system architectures, agents often improvise unverified solutions instead of grounding decisions in proven prior art. This Skill enforces a systematic search-audit-adapt procedure so implementations are anchored to production-tested code, official standards, or peer-reviewed literature rather than internal confidence. ## Core Features & Use Cases - Tiered Search Hierarchy: Searches Tier 1 permissively licensed production code (MIT, Apache-2.0, BSD), then Tier 2 official standards and RFCs (extracting RFC 2119 normative constraints), then Tier 3 academic literature and formal models (arXiv, TLA+), halting if no grounding exists. - Workspace Hygiene for Cloning: Mandates shallow clones (--depth 1), sparse checkouts, and an isolated .prior_art_cache/ directory, with a purge gate requiring rm -rf .prior_art_cache/ before any commit boundary. - Invariant Auditing: Blocks direct copy-paste by auditing imported logic against spatial simplicity (Hickey audit) and temporal volatility (Lowy audit), and records findings with licenses, citations, and extracted invariants in the active sketch ledger. - Use Case: When implementing a consensus protocol, the Skill first audits an Apache-2.0 reference implementation, maps RFC normative constraints, extracts safety invariants from a TLA+ specification, and documents how each invariant adapts to the target codebase. ## Quick Start Use the prior-art skill to research existing implementations and standards before designing this distributed rate limiter.