AgentSkillsCN

generate-linkedin-post

生成一篇关于已完成工作包的 LinkedIn 帖子，其中包含完整的故事线、详细的实施过程，以及引导用户参与互动的行动号召。

SKILL.md
--- frontmatter
name: generate-linkedin-post
description: Generate a LinkedIn post about a completed work package with story arc, implementation details, and engagement CTA
user-invocable: true

Skill: Generate LinkedIn Post

Generate a LinkedIn post about a completed work package from the AI Engineering Monorepo project.

When to Use

  • •After completing a work package (triggered by the WP Completion Checklist in CLAUDE.md)
  • •When the user explicitly asks for a LinkedIn post about their progress or interesting concept to be shared

Inputs

Before generating, read:

  • •PROGRESS.md for WP objectives, dates, and notes
  • •.claude/learning-progress.md for skills learned
  • •.claude/learning-context.md for developer profile and session concepts

Post Structure (Story Arc)

  1. •Hook (first 150 characters) - compelling opening that makes people click "see more"
  2. •Setup (1-2 sentences) - what project, what phase, what was the goal
  3. •Implementation (3-5 bullet points) - what was built, with specific technologies
  4. •Lesson (2-3 sentences) - key insight or takeaway from the process
  5. •CTA - closing question to drive engagement

Format Rules

  • •Length: 1,200-1,600 characters (LinkedIn algorithm sweet spot)
  • •Line breaks: Single-line paragraphs with blank lines between (LinkedIn collapses long blocks)
  • •Bullet points: Use arrow symbols (->) for visual scanning
  • •Emojis: Maximum 1-2 strategic emojis, placed at section transitions
  • •Hashtags: 3-5 relevant hashtags on their own line at the end
  • •No links in body: LinkedIn suppresses posts with links; add link in comments instead (note this in the output file)
  • •Language: English

Tone Guidelines

  • •First person, professional but approachable
  • •Frame as upskilling and staying current, not "learning from scratch"
  • •The author is an experienced Python developer expanding into new tools and patterns
  • •Be specific about technologies and what was built
  • •Mention challenges faced and how they were overcome
  • •Mention Claude Code as an agentic coding tool being used in the process and that I am better understanding advanced usage skills, like skills, mcps, prompt engineering, Programmatic Tool Calling (PTC) and Tool Search.
  • •Be transparent about "vibe coding" for TypeScript/UI parts (comfortable with Python, using AI to expand into TypeScript for more robust frontends than Streamlit)
  • •Avoid buzzwords and generic statements
  • •No AI-generated feel; write as if sharing with engineering peers
  • •NEVER use phrases like "I decided to learn X" or "I'm learning X" - instead frame as "building", "expanding into", "updating my stack with"

Template

code
[Hook - compelling first line that creates curiosity, under 150 chars]

[1-2 sentences of context: what project, what phase, what motivated this]

Here's what I built:

-> [Specific thing 1 with technology name]
-> [Specific thing 2 with technology name]
-> [Specific thing 3 with technology name]
-> [Optional: Specific thing 4]

[Key insight or lesson learned - what surprised you, what you'd do differently, or what clicked]

[CTA question to audience - specific enough to invite real answers]

#hashtag1 #hashtag2 #hashtag3 #hashtag4

Hook Examples (for this project)

  • •"I just built a CI pipeline that runs 4 checks in parallel. Here's what broke along the way."
  • •"Most people learn Python from tutorials. I'm building a production monorepo from scratch."
  • •"13 linting errors. 3 type failures. 1 working pipeline."
  • •"I decided to learn AI Engineering by building three real projects simultaneously."
  • •"Setting up a Python + TypeScript monorepo taught me more than any course."

Hashtag Pool

Pick 3-5 from this baseline - identify skills/frameworks from the WP to add as well: #AIEngineering #Python #DataEngineering #LangChain #MachineLearning #DevOps #CICD #GitHubActions #OpenSource #LearningInPublic #BuildInPublic #SoftwareEngineering #MonorepoArchitecture #TypeScript #MLOps

Output Format

Save each post to .claude/linkedin-posts/WP-XXX-short-title.md with this structure:

markdown
# LinkedIn Post: WP-XXX - Title

## Post

[The full post text, ready to copy-paste into LinkedIn]

---

## Notes

**Suggested first comment:** "Project repo: [link]. Built with [key technologies]."

**Diagram suggestion:** Generate a PROMPT to pass to Gemini Banana to generate a visual diagram of [specific topic from this WP] to accompany this post.

**Best posting time:** Tuesday-Thursday, 8-10am local time

**Character count:** [count]