LazyIsEfficient Expert Panel
General-purpose scoring and iterative improvement engine. Auto-assembles the right experts for whatever is being evaluated, scores it, and loops until 90+.
Core rules
- Intake: collect content, content type, offer context, variants, and source skill — full procedure in
references/procedure-steps.mdStep 1. - Auto-assemble 7–10 experts: start from
experts/pre-built panels, add 1–3 domain experts, always include AI Writing Detector (1.5x weight) and Brand Voice Match. - Select scoring rubric from
scoring-rubrics/by content type; read the file for criteria. - Score recursively until 90+ aggregate (max 3 rounds). Humanizer weighted 1.5x. Show all rounds in output — the iteration trail is the value.
- Check
references/patterns.mdat every round start and dock points for known-bad patterns before expert scoring. - When scoring another skill's output, generate a Source Improvement Brief (Step 6).
- On user rejection of 90+ content, capture the reason and append to
references/patterns.md.
References
- references/procedure-steps.md — full 7-step procedure: intake, panel assembly, rubric selection, scoring loop, output format, feedback-to-source, pattern learning
- references/expert-assembly.md — domain-expert examples for auto-assembly of unfamiliar panels
- references/patterns.md — learned rejection patterns; read every run
- experts/humanizer.md — AI writing detection rubric (24 patterns); always run
- experts/ — pre-built panels: humanizer, instagram, linkedin, newsletter, podcast-quotes, recruiting, seo-strategy, x-articles, youtube-shorts
- scoring-rubrics/ — content-quality, conversion-quality, evaluation-quality, strategic-quality, visual-quality
Related skills
- autoresearch — pre-launch variant generation + multi-round optimization of conversion copy; run before content-ops's final gate
- conversion-ops — post-publish conversion layer; run after content-ops quality gate
- adversarial-claims-reviewer — judges whether formal/technical claims are true; content-ops judges whether the prose is good
- content-pipeline — script-driven content production (RSS quote mining, video-clip discovery, repurposing, batch draft gating); reuses this skill's
experts/panels in its transform stage