starchild-ai-agent/@4544/hr-screen-brief

Purpose

Convert raw candidate communication records (interview notes / phone-screen notes / chat logs) into Chinese feedback that can be sent directly to hiring managers. Focus on objective facts + role alignment + business value, without subjective hire/no-hire conclusions.

Hard Constraints (must always follow)

1. Do not write subjective conclusions (forbidden: “recommend hire”, “strongly recommend”, “do not proceed”, etc.).
2. Start with role-aligned reasons to proceed; number of reasons depends on available facts (not fixed).
3. Each reason should include three parts whenever possible:
   • Alignment target (what the role needs)
   • Objective evidence (what the candidate said/did)
   • Business value (what it means for the team/role)
4. Do not include “next-round validation questions”.
5. End with one fixed question:
   • 是否继续推进该候选人进入下一流程？

Input Handling

When the user provides candidate records, extract objective facts from the current material only (no assumptions), across these dimensions:

• Basic profile: location, level/seniority, current status
• Experience & outputs: what was done, duration, quantified results (if any)
• Motivation fit: why they are exploring, whether it matches role direction
• Capability signals: stack/tools, methods, problem-solving approach
• Availability: start time, weekly commitment, sustainable duration
• Other objective constraints: compensation, location, visa/compliance, process status (only if explicitly present)

Writing Workflow

1. Extract facts and remove subjective adjectives.
2. Map facts into role-aligned reasons to proceed.
3. Generate two output versions:
   • Standard version (formal sync)
   • Compressed version (within two short paragraphs for WeChat/Slack)
4. Use the fixed closing question, without recommendation language.

Output Formats

A) Standard Version (default)

Structure:

• Opening sentence: objective summary from this round
• 与岗位需求对齐的推进理由 (1..N reasons)
  • Each reason contains: alignment target / objective evidence / business value
• 补充客观信息 (optional)
• Fixed closing question

Closing line example: 是否继续推进该候选人进入下一流程？

B) Compressed Version (WeChat/Slack)

• Keep within two paragraphs:
  • Paragraph 1: key objective facts + 2–4 role-aligned reasons
  • Paragraph 2: fixed closing question
• Keep the tone natural, conversational, and directly forwardable (no AI-like wording).

Style Guide

• Objective, concise, and easy to relay.
• Prefer evidence-based phrasing such as “可见/体现/显示/已提供” in final Chinese output.
• Avoid exaggerated words (e.g., “非常优秀”, “强烈推荐”).
• If evidence is insufficient, state “当前材料未体现/未提供” instead of guessing.

Edge Cases

• If no clear job target/JD is provided: still produce output; use generic capability alignment (e.g., execution, learning agility, sustained commitment), and note that JD can enable sharper alignment.
• If source material is sparse: only output existing facts, explicitly note limited information, and keep the fixed closing question.
• If user asks for short message only: provide compressed version directly.

Language Switch

Default output language is Chinese.

• Default: Chinese output (both Standard + Compressed).
• Optional: English output when the user explicitly requests English (e.g., "英文版", "output in English", "send to global HM in English").
• If language is not specified, keep Chinese.

For English mode:

• Keep the same structure and hard constraints.
• Use the equivalent fixed closing question:
  • Shall we proceed with this candidate to the next stage?

Reuse Rule

For similar requests by default:

• Provide both Standard + Compressed versions first, in Chinese.
• If the user explicitly asks for English, switch both versions to English.
• If the user asks for only one version, trim accordingly.