interview-collector

# Interview Collector Use this skill when the user asks to collect, summarize, deduplicate, or update AI Agent / LLM / RAG interview experiences. ## Rules - Use only public accessible sources unless the user logs in through their browser. - Do not request, expose, store, or commit cookies, tokens, or credentials. - Do not bypass paywalls, login walls, or anti-bot controls. - Summarize sources; do not copy long external text. - Preserve source attribution. ## Workflow 1. Search: 牛客、小红书、知乎、CSDN、博客园、掘金、GitHub、RSS. 2. Read public pages with Jina Reader or platform tools. 3. Extract structured fields. 4. Deduplicate. 5. Score quality. 6. Write candidates to `data/interviews.json`. 7. Generate Markdown summaries and run the static site build. ## Commands ```bash python3 scripts/collect_interviews.py doctor python3 scripts/collect_interviews.py search --query "AI Agent 大模型 面经 2026" agent-reach doctor --json mcporter call 'exa.web_search_exa(query: "AI Agent 大模型 面经 2026", numResults: 10)' curl -s "https://r.jina.ai/https://example.com/page" opencli xiaohongshu search "AI Agent 面经" -f yaml --window background gh search repos "AI Agent interview" --limit 10 --json fullName,description,stargazersCount,url,updatedAt python3 -m json.tool data/interviews.json python3 scripts/build_site.py ``` ## Data model ```json { "id": "stable-kebab-id", "platform": "牛客", "title": "面经标题", "company": "字节跳动", "role": "AI Agent开发", "published_at": "2026-05-20", "source_url": "https://...", "source_note_id": null, "source_lookup": null, "score": 5, "topics": ["Agent", "RAG", "MCP"], "summary": "一到两句摘要。" } ``` For Xiaohongshu, avoid unstable direct links in public Markdown. Use `source_note_id` and `source_lookup`.