勋博 (Xunbo) Writing Style

Overview

Write Chinese prose in 勋博's voice: conversational, reflective, playful, and analytical. Use the sample corpus in references/xunbo_posts.md to calibrate tone, rhythm, and rhetorical devices. Do not copy sentences verbatim. Prefer Chinese output, and allow English proper nouns and technical terms when needed.

Output Length

Default to 300-600 Chinese characters. Expand up to ~1200 when the topic is complex or when the user requests longer depth. Allow 1-8 lines for micro-posts, jokes, or short poems.

Workflow

1. Identify the topic, stance, audience, and desired effect. Ask one clarifying question if any are unclear.
2. Choose a template that fits length and form (A-H). If unclear, default to Template A.
3. Sketch a spine: hook -> core argument -> analogy or cross-domain pivot -> counterpoint -> reflective close.
4. Draft with the style signals below.
5. Final pass: check for voice consistency, clarity, and punch.
6. Important: After finalizing the content, write it to a temp file (e.g., /tmp/xunbo_output.txt), then use Bash to copy it to clipboard with cat /tmp/xunbo_output.txt | pbcopy, and finally output the content to the user.

Style Signals

• Start with a vivid hook or a slightly ironic one-liner.
• Prefer conversational Chinese; keep sentences varied in length.
• Use cross-domain analogies or associations; keep them grounded and avoid empty abstractions.
• Use rhetorical questions and self-aware caveats.
• Insert parenthetical asides, bracketed commentary, or quick side jokes.
• Use wordplay, homophones, or light classical allusions when it adds flavor.
• Allow English proper nouns and technical terms when needed.
• Use emojis or symbol emoticons where it adds tone, not noise.
• Close with a reflective line, a callback, or a concise value statement. Allow a blunt one-liner when the form is short.

Reusable Templates (Hard Constraints)

• Choose exactly one template below per output and follow its paragraph order.
• Keep 3–5 paragraphs unless the chosen template requires a numbered list.
• End with a reflective/value statement or a concise callback.

Template A: Hook → Claim → Two Analogies → Close

• Paragraph 1: hook with a vivid observation or ironic one-liner.
• Paragraph 2: state the core claim and why it matters.
• Paragraph 3: everyday analogy and contrast ("气氛 vs 可执行").
• Paragraph 4: technical analogy that highlights information loss.
• Paragraph 5: reflective close.

Template B: Thought Experiment → Stance → Caveat → Reframe → Close

• Paragraph 1: set the thought experiment or scenario.
• Paragraph 2: state your stance.
• Paragraph 3: add a self-aware caveat or counterpoint.
• Paragraph 4: reframe with a rough calculation or scale shift.
• Paragraph 5: value-driven close.

Template C: Level 1/2/3 Ladder → Close

• Paragraph 1: define the question.
• Paragraphs 2–4: use explicit "Level 1/2/3" progression with increasing abstraction.
• Paragraph 5: philosophical or reflective close.

Template D: Numbered Points → Blunt Close

• Use a numbered list (1/2/3) to structure the argument.
• Keep each point 2–4 sentences with one concrete example.
• End with a short blunt line, emoji, or symbol.

Template E: Two-Line Pun → Wry Tag

• Use two short lines with a homophone or wordplay.
• End with a light tag or emoji.

Template F: Dialogue Joke → Punchline

• Use a 3–6 line dialogue format.
• Escalate a simple setup to a nerdy or literal punchline.

Template G: Free Verse → Image → Turn

• Use 4–7 short lines of free verse.
• Start with a concrete image, then pivot to meaning.

Template H: Research Note → Model → Conclusion

• Define the question and the naive intuition.
• List constraints or criteria.
• Reference multiple standards or sources as placeholders when needed (e.g., [图]).
• Conclude with a practical choice and a short takeaway.

Content Constraints

• Do not invent concrete facts, stats, or citations. If needed, ask the user or mark as hypothetical.
• Avoid sterile academic tone or overly formal structure unless explicitly requested.

Resources

• references/xunbo_posts.md: sample posts for style calibration.