Zhang Zhen Recruiting Assistant

Problem

Recruiting workflows often stall at setup: missing Feishu table permissions, failed lark-cli authentication, lost browser logins, or a private schema being forced onto another recruiter's table. This skill makes environment readiness a gate before resume work begins.

How it works

• Environment doctor: run env_doctor.py on each new machine; if status is not ready, follow need_auth_login, need_admin_permission, need_install_tools, or similar member_action guidance instead of experimenting.
• Feishu table writing: read real headers, map fields by meaning, avoid unrelated columns, highlight only changed cells, and preserve existing phone, email, WeChat, and profile links unless overwritten intentionally.
• Resume intake: local resumes default to extraction and entry only; web resumes are processed one by one, with open, read, write, and verify steps.
• Structured tagging: build one-level-to-three-level labels for software/AI or industry roles, such as Software-AI-Large Model or Finance-Research-Quant Research.
• Public search: use GitHub, Google Scholar, OpenReview, arXiv, and personal pages for technical candidates; for other roles, search name, company, title, city, award, project, patent, article, and conference signals.

Boundaries

It assumes a Feishu table link, logged-in browser state, and basic automation are available. If credentials, permissions, or host execution are blocked, stop at the diagnosis status rather than falling back to bot APIs, browser DOM scraping, CSV export, or link-sharing changes.