HR Resume Screener

Problem: repetitive resume triage and scattered outputs

In hiring workflows, JDs, PDF or Word resumes, Excel lists, and chat attachments are scattered across entry points. Manual comparison can miss key match points, the same candidate may be analyzed repeatedly, and interview preparation still needs quickly generated talking points. This skill turns JD-to-resume matching into inspectable steps: parse inputs, run AI matching against references/resume_analysis_prompt.md, then export a report and create a candidate archive package.

How it works and boundaries

• Input parsing: supports PDF, Word (.docx), Excel, images, and text files, triggered by a single file, a resume folder, or folder monitoring.
• Deduplication: uses candidate_registry.json and treats a name, phone number, or email match as the same candidate; duplicates can be reviewed, reanalyzed, or skipped.
• Output routing: writes results to configured channels such as chat, json, excel, word, kdocs, tencent_docs, wechat_push, or api_output, and can generate candidate archive packages on the selected platform.

Note that it depends on the configured storage platform and channel adapters. If the JD knowledge base is empty, a channel is unavailable, or a file exceeds limits, it may prompt for setup or fall back to default outputs. It is suitable for initial screening, match notes, and interview question lists that need an auditable record, especially for batch resume folders, but it does not replace HR judgment for final hiring decisions.