Diagnose built UI, workflow, spec, or implementation issues with criteria and options.
npx skills add https://github.com/Undertone0809/rudder --skill build-advisor
Please help me install this Agent Skill. Skill: build-advisor Source: https://github.com/Undertone0809/rudder/tree/main/.agents/skills/build-advisor Command: npx skills add https://github.com/Undertone0809/rudder --skill build-advisor
Build-advisor helps when something you built (UI, workflow, spec, or implementation) feels wrong or weak, but you cannot clearly articulate what is broken or what the right next move should be.
Ask build-advisor to analyze your build using evidence you have (screenshots, transcript, traces, benchmarks, or artifacts) and produce an evaluation-ready critique with criteria, options, and a recommended next move.
High-intent search queries and answers about installing and using this skill.
To diagnose a UI workflow that feels wrong, use first-principles scenario analysis to convert vague discomfort into a clearer problem statement, identify the affected layer, and map requirements. This process reframes fuzzy dissatisfaction into an explicit evaluation rubric.
The best way to evaluate a product implementation is establishing criteria-first evaluation. Define explicit evaluation criteria and a grounded rubric to judge whether the next iteration is actually better, ensuring your post-build critique is objective rather than subjective.
You can get actionable options for a broken design spec by generating a decision-ready recommendation. This approach provides 2-3 realistic options and a concrete next move, routing toward specialized follow-up skills if deeper diagnosis is required.
To determine if a build issue is product framing or engineering shape, perform layer identification. This diagnosis evaluates whether the problem stems from product framing, UX, engineering shape, workflow process, or observability evidence.
For an effective UI critique, provide evidence intake such as screenshots, transcripts, traces, benchmarks, or artifacts. Supplying these materials allows the analysis to ground its evaluation criteria and corner-case coverage in real implementation data.
readycheck-devExtract user observations from ADA transcripts and correlate them with trace events and screenshots.
ManzuetiAutomate sequential CEO, design, engineering, and DX reviews for software projects.
lguibrInstrument applications to generate LangSmith traces for LLM observability.
SufficientDaikonReview design artifacts using heuristic evaluation and WCAG 2.1 AA criteria.
