Interview Question Generator & Evaluator
Two-phase workflow for WePlay activity operations (活动运营) interviews using Feishu docs.
Phase 1: Resume → Interview Question Document
Step 1: Read the Resume
If the resume is a PDF attachment, render each page as an image (/tmp/resume_p{n}.png) using PyMuPDF and read them visually:
import fitz
doc = fitz.open("/path/to/resume.pdf")
for i, page in enumerate(doc):
page.get_pixmap(matrix=fitz.Matrix(1.5, 1.5)).save(f"/tmp/resume_p{i+1}.png")
Extract key info: work experience, skills, education, highlights.
Step 2: Read WePlay Product Context
Before generating questions, fetch the WePlay product framework doc to understand product positioning:
Step 3: Generate Interview Questions
Structure the document into these sections. See references/question-template.md for the full question template and scoring rubrics.
Document sections:
- 破冰与自我介绍 (2 questions)
- 结合简历的深挖问题 (4–6 questions, grouped by employer)
- 活动运营能力考察 (4 questions: scenario planning, data, cross-team collaboration)
- 日语与本地化能力 (3 questions, tailored to Japanese market)
- WePlay 产品体验问题 (5 questions — require candidate to pre-download WePlay)
- 价值观与潜力考察 (4 questions including open Q&A)
- 日本語口頭試問 (6 questions, all in Japanese — no Chinese)
Tailor questions to the specific candidate's background. Reference their actual projects, metrics, and employers by name.
Step 4: Create Feishu Document
Use feishu_bot_doc.mjs to create the document:
cat /tmp/interview_questions.md | node scripts/feishu_bot_doc.mjs create \
--title "【AI生成】{候选人姓名} {岗位} 面试题集" \
--stdin \
--folder AZ3nfFtial4bHTdOFahcdcfxnub \
--collaborator ou_8b357150cff930fca19a733461a32526
Reply with the document URL. Tell the user to send the interview transcript when ready.
Phase 2: Interview Transcript → Evaluation
Step 1: Read the Transcript
Accept the transcript as:
- A Feishu doc link → use
feishu_doc read action
- A pasted text block → read directly
Step 2: Write Evaluation
Append the evaluation to the existing interview question document (not a new doc). Use feishu_doc append action on the same doc_token.
See references/evaluation-template.md for the full evaluation structure and scoring rubrics.
Evaluation structure:
- 总体印象 (1–2 sentences, overall rating: 优秀/良好/中等/中等偏下/不建议录用)
- 各维度评价 with ⭐ ratings (1–5 stars each):
- 过往经验匹配度
- 活动策划思维
- 数据分析能力
- 产品认知与洞察
- 日本市场理解
- 表达与沟通
- 亮点 (bullet list)
- 主要风险 (bullet list)
- 结论 (录用 / 待定 / 不建议录用, with reasoning)
Be specific: quote actual interview moments, not generic observations.
Notes
- Japanese oral exam questions (Section 7) must be written entirely in Japanese — no Chinese text.
- Always add
【AI生成】 prefix to document titles.
- Default save folder:
AZ3nfFtial4bHTdOFahcdcfxnub
- Default collaborator:
ou_8b357150cff930fca19a733461a32526 (吴柏庆)
- If the interview transcript doc is auto-generated by Feishu (智能纪要), the bot has no write permission — append to the question doc instead.
