yosishe
UserAudit AI skills and prompts for what they actually cost per trigger — and for reference files your model can never reach. Never touches safety text; will not print a number it cannot trace to data.
Categories
Indexed Skills (3)
token-efficient-skill-optimizer
Audit and optimize an existing AI skill, system prompt, agent instruction set, or workflow for token/cost efficiency WITHOUT degrading quality or safety — evidence-backed rules, honest measurement (measured/estimated/projected labels enforced by a validator), reviewable diffs, before/after benchmarks. Use when the user wants to cut a skill's or prompt's token cost, context footprint, or API spend; audit why an agent is expensive; validate someone's claimed token savings; or batch-audit a skills directory. Triggers: "optimize this skill/prompt", "cut token costs", "why is this agent so expensive", "audit my skill", "is this optimization real", "לייעל את הסקיל", "לחסוך טוקנים", "כמה עולה הסקיל הזה", "בדוק את החיסכון". Do NOT use for one-off prompt-wording help that won't be saved as a reusable artifact (answer directly), for authoring a brand-new skill from scratch (use skill-creator / token-efficient-skill-builder), or for optimizing a skill whose purpose is harmful — refuse those.
bundle-skill
Fixture. Use when testing compiled-bundle handling. Do not use for anything real.
convention-skill
Fixture. Use when testing convention-based reachability. Do not use for anything real.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.