← ClaudeAtlas

loopy-theorylisted

Use when a project, product, workflow, codebase, or document set should be treated as imperfect evidence of a simpler underlying theory; infer the clearer principle that current implementation traces most strongly suggest, test it against gaps, and produce improvement candidates without directly changing implementation.
phykn/loopy · ★ 0 · AI & Automation · score 70
Install: claude install-skill phykn/loopy
# Loopy Theory Use this skill to apply the Loopy counterexample loop to project theory. Loopy Theory treats the current project as imperfect observations. The goal is not to guess prior motives or justify the current implementation. The goal is to infer the simpler and more coherent principle that the current evidence most strongly suggests, then find where the project falls short of that principle. ```text Observe -> Infer -> Recheck -> Challenge -> Refine -> Imply -> Decide -> Route ``` ## Invocation Contract When invoked, start by establishing a compact Done Contract before inferring a theory. Do not mark the theory stable or open final candidates until the theory Done Gate has run: 1. name the strongest gap considered; 2. name one competing in-scope theory or gap that could overturn the result; 3. either refine, or explain why it no longer changes the theory, candidates, or next route. ## Done Contract Before analyzing project evidence, establish a Done Contract. The contract must name: - `Target`: what the theory should explain or decide, and what evidence must not be lost; - `Non-goals`: implementation changes or claims outside the current evidence; - `Checks`: evidence criteria, gap checks, and final review criteria; - `Counterexamples`: gaps that would force another loop; - `Stop condition`: what proves no important in-scope gap remains. Keep it compact, but do not start theorizing until the contract is clear enough to test. ## Verification Policy Before