Youhuo Resume Assistant

Problem

Resume reviews often mix campus, experienced, and gig candidates into one scoring model, output only a generic total score, and compress technical stacks, metrics, and architecture details when forcing a one-page export. Users asking for application channels are also often asked to submit a resume before getting a useful official portal. This skill turns those constraints into explicit execution rules, reducing the risk that a model looks polished while inventing unconfirmed facts.

How It Works

• Track separation: Uses separate scoring weights and rewriting standards for campus, experienced, and gig roles, so campus candidates are not judged by senior-level outcome metrics.
• Dimensional scoring: Diagnosis outputs each dimension, weight, score, and a one-sentence reason instead of only X/10; major rewrites trigger rescoring and a before/after comparison.
• STAR and numeric guardrails: Rewrites extract facts from the original text first. New numbers, company names, titles, or role upgrades that the user has not confirmed go into a pending pool rather than the final export.
• Export fidelity: Content is written to structured JSON, rendered to HTML, and optionally exported to PDF or Word. Long bullets are split rather than truncated, and tech stacks must remain intact.
• Portal navigation: Users can ask for official job portals by role or company without providing a resume first. The skill returns links and steps, but does not log in, fill forms, send emails, or batch apply.

Boundaries

It is suitable for resume diagnosis, STAR rewriting, JD matching, export, and official portal navigation. It is not suitable for automated third-party job boards. If a portal link fails, the skill notes that the database may be incomplete or outdated and recommends manual website search. Full multi-turn interview simulation is handled by a neighboring skill; this one focuses on question prediction and material preparation.