Claycui828/ASu-resume-skills

What it solves

Writing a resume often means balancing between making real achievements sound impressive and avoiding exaggeration. This project provides two Codex skills that work together: one turns your true experience into a high-density, well-structured resume without inventing facts, and the other checks every strong claim against verifiable evidence so you know exactly what is supported and what is not.

How it works

The first skill ($asu-resume-skill) reads your raw materials – existing resumes, Feishu docs, screenshots, project files – and produces a source-mapped JSON, a self-contained HTML file, an A4 PDF, and social-media copy. It verifies school titles using official sources, breaks real projects into technical chains (Data/Model/Training/Serving/Agent/Eval), and assigns clear ownership levels (module Owner, 0→1, architecture evolution) without upgrading participation into full ownership. The second skill ($asu-resume-audit-skill) takes a resume and splits every claim into atomic statements that are marked as confirmed, partially confirmed, contradicted, unsupported, or unverifiable. It looks at roles, scope expansion, metric definitions, and tense boundaries (delivered vs. planned). The two skills are meant to be used in sequence: generate, audit, fix weak points, then regenerate the final output.

Who it’s for

This is for developers, tech leads, and anyone preparing promotion materials or job applications in Chinese tech companies. It’s also useful for hiring managers or reviewers who want to verify the strength of a candidate’s claims. Both skills run inside Codex as plugins, so you need a Codex-compatible environment.

Highlights

• Two complementary skills: resume generation and resume auditing
• Outputs: self-contained HTML, A4 PDF, JSON data, and social copy (for generation); evidence matrix and audit report (for auditing)
• Strict fact boundary: no fake schools, projects, metrics, or outcomes can pass through
• Verifies school titles via official channels
• Decomposes claims into atomic, evidence-checkable statements
• Distinguishes between “0→1” as new service, new training pipeline, or new strategy node
• Does not assume dishonesty; only reports what evidence supports
• Includes complete example HTML/PDF files and JSON data
• MIT licensed