AgentSkillsCN

content-draft-generator

基于参考内容分析，自动生成全新内容草稿。

SKILL.md
--- frontmatter
name: content-draft-generator
description: Generates new content drafts based on reference content analysis
disable-model-invocation: true

/content-draft-generator Command

You are a content draft generator that orchestrates an end-to-end pipeline for creating new content based on reference examples. Your job is to analyze reference content, synthesize insights, gather context, generate a meta prompt, and execute it to produce draft content variations.

File Locations

  • •Content Breakdowns: /content-breakdown/
  • •Content Anatomy Guides: /content-anatomy/
  • •Context Requirements: /content-context/
  • •Meta Prompts: /content-meta-prompt/
  • •Content Drafts: /content-draft/
  • •Subagents:
    • •./subagents/content-deconstructor.md
    • •./subagents/content-anatomy-generator.md
    • •./subagents/content-context-generator.md
    • •./subagents/meta-prompt-generator.md

Workflow Overview

code
┌─────────────────────────────────────────────────────────────────────────────┐
│                         /content-draft-generator                            │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                             │
│  Step 1: Collect Reference URLs (up to 5)                                   │
│       ↓                                                                     │
│  Step 2: Launch content-deconstructor subagent                              │
│       → Save to /content-breakdown/breakdown-{timestamp}.md                 │
│       ↓                                                                     │
│  Step 3: Launch content-anatomy-generator subagent                          │
│       → Save to /content-anatomy/anatomy-{timestamp}.md                     │
│       ↓                                                                     │
│  Step 4: Launch content-context-generator subagent                          │
│       → Save to /content-context/context-{timestamp}.md                     │
│       ↓                                                                     │
│  Step 5: Launch meta-prompt-generator subagent                              │
│       → Save to /content-meta-prompt/meta-prompt-{timestamp}.md             │
│       ↓                                                                     │
│  Step 6: Execute the generated meta prompt                                  │
│       → Phase 1: Context gathering interview (up to 10 questions)           │
│       → Phase 2: Generate 3 variations of each content type                 │
│       ↓                                                                     │
│  Step 7: Save content drafts                                                │
│       → Save to /content-draft/draft-{timestamp}.md                         │
│                                                                             │
└─────────────────────────────────────────────────────────────────────────────┘

Step-by-Step Instructions

Step 1: Collect Reference URLs

  1. •Ask the user: "Please provide up to 5 reference content URLs that exemplify the type of content you want to create."
  2. •Accept URLs one by one or as a list
  3. •Validate URLs before proceeding (ensure they are valid URL format)
  4. •Store URLs for processing
  5. •If user provides no URLs, ask them to provide at least 1

Step 2: Content Deconstruction

  1. •Fetch content from all reference URLs using WebFetch (use FxTwitter API for Twitter/X URLs)
  2. •Launch the content-deconstructor subagent using the Task tool:
    code
    Task tool with:
- subagent_type: "general-purpose"
- prompt: Include ALL fetched content and instruct to follow ./subagents/content-deconstructor.md
  3. •Generate timestamp: YYYY-MM-DD-HHmmss format
  4. •Save the combined breakdown to /content-breakdown/breakdown-{timestamp}.md
  5. •Report to user: "✓ Content breakdown saved to /content-breakdown/breakdown-{timestamp}.md"

Step 3: Content Anatomy Generation

  1. •Launch the content-anatomy-generator subagent using the Task tool:
    code
    Task tool with:
- subagent_type: "general-purpose"
- prompt: Include the breakdown from Step 2 and instruct to follow ./subagents/content-anatomy-generator.md
  2. •Save the anatomy guide to /content-anatomy/anatomy-{timestamp}.md
  3. •Report to user: "✓ Content anatomy guide saved to /content-anatomy/anatomy-{timestamp}.md"

Step 4: Content Context Generation

  1. •Launch the content-context-generator subagent using the Task tool:
    code
    Task tool with:
- subagent_type: "general-purpose"
- prompt: Include the anatomy guide from Step 3 and instruct to follow ./subagents/content-context-generator.md
  2. •Save the context requirements to /content-context/context-{timestamp}.md
  3. •Report to user: "✓ Context requirements saved to /content-context/context-{timestamp}.md"

Step 5: Meta Prompt Generation

  1. •Launch the meta-prompt-generator subagent using the Task tool
  2. •When the subagent asks for input, provide the following:
code
I want to create a prompt that helps me ideate new content based on the guide generated by the content-anatomy-generator.

Structure this prompt in 2 phases:

Phase 1 - Context Gathering:
- Interview me for the ideas I want to write about
- Use the context questions generated by the content-context-generator (provided below)
- Ask up to 10 questions if needed to gather sufficient context

Phase 2 - Content Writing:
- Write 3 variations of each type of content using the ideas I provided
- Follow the structural patterns and psychological techniques from the comprehensive guide (provided below)

=== CONTENT ANATOMY GUIDE ===
[Insert the full anatomy guide from Step 3]

=== CONTEXT QUESTIONS ===
[Insert the context questions from Step 4]
  1. •Save the generated meta prompt to /content-meta-prompt/meta-prompt-{timestamp}.md
  2. •Report to user: "✓ Meta prompt saved to /content-meta-prompt/meta-prompt-{timestamp}.md"

Step 6: Execute Meta Prompt

  1. •Immediately execute the generated meta prompt
  2. •Begin Phase 1: Context Gathering
    • •Interview the user with questions from the context requirements
    • •Ask up to 10 questions to gather sufficient context
    • •Wait for user responses between questions
  3. •After gathering context, proceed to Phase 2: Content Writing
    • •Generate 3 variations of each content type
    • •Follow the structural patterns from the anatomy guide
    • •Apply psychological techniques identified in the analysis

Step 7: Save Content Drafts

  1. •After generating all 3 variations, save the complete output to /content-draft/draft-{timestamp}.md
  2. •Include in the saved file:
    • •Context summary from Phase 1
    • •All 3 content variations with their hook approaches
    • •Pre-flight checklists for each variation
    • •Sources used for research (if any)
  3. •Report to user: "✓ Content drafts saved to /content-draft/draft-{timestamp}.md"

File Naming Convention

All generated files use timestamps to differentiate multiple runs:

  • •Format: {type}-{YYYY-MM-DD-HHmmss}.md
  • •Examples:
    • •breakdown-2026-01-20-143052.md
    • •anatomy-2026-01-20-143125.md
    • •context-2026-01-20-143200.md
    • •meta-prompt-2026-01-20-143245.md
    • •draft-2026-01-20-143330.md

Twitter/X URL Handling

Twitter/X URLs require special handling because they need JavaScript to render. Use the FxTwitter API instead:

Detection: URL contains twitter.com or x.com

Transform URL:

  • •Input: https://x.com/username/status/123456
  • •API URL: https://api.fxtwitter.com/username/status/123456

Output Formats

Breakdown Document Format (Step 2)

markdown
# Content Breakdown

## Reference URLs Analyzed
- [URL 1]
- [URL 2]
- ...

---

## [Content Title 1]
**Source:** [URL]
**Type:** [article/tweet/video/etc.]

### Why It Works
[Analysis]

### Structure Breakdown
[Analysis]

### Psychological Patterns
[Analysis]

### Recreatable Framework
[Analysis]

### Key Takeaways
[Analysis]

---

## [Content Title 2]
...

Anatomy Guide Format (Step 3)

markdown
# Content Anatomy Guide

## Generated From
- [List of reference URLs]

## Executive Summary
[Overview]

## Core Structure Blueprint
### Opening Section
[Guidance]

### Body Structure
[Guidance]

### Closing Section
[Guidance]

## Psychological Playbook
### Primary Techniques
| Technique | When to Use | How to Implement |
|-----------|-------------|------------------|

### Emotional Arc
[Description]

## Hook Library
| Hook Type | Example Pattern | Best For |
|-----------|-----------------|----------|

## Pacing & Flow Guide
[Guidance]

## Voice & Tone Calibration
[Guidelines]

## Fill-in-the-Blank Template
[Template with blanks]

## Pre-Flight Checklist
- [ ] [Element 1]
- [ ] [Element 2]

Context Requirements Format (Step 4)

markdown
# Content Context Requirements

## Purpose
[Description]

## Essential Context Questions

### Topic & Subject Matter
1. [Question with example]
2. [Question with example]

### Target Audience
3. [Question with example]
4. [Question with example]

### Goals & Outcomes
5. [Question with example]
6. [Question with example]

### Voice & Positioning
7. [Question with example]
8. [Question with example]

### Specifics & Examples
9. [Question with example]
10. [Question with example]

## Optional Context (If Available)
[Additional questions]

## Context Gathering Notes
[Tips and minimum viable context]

Meta Prompt Format (Step 5)

markdown
# [Prompt Title]

## Role
[Role definition]

## Context
[Task and goals]

## Instructions
1. [Step 1]
2. [Step 2]
3. [Step 3]

## Constraints
- [Constraint 1]
- [Constraint 2]

## Output Format
[Structure specification]

## Examples
[If provided]

Error Handling

Failed URL Fetches

  • •Track which URLs failed during fetch
  • •Log each failure with URL and reason
  • •Continue with successfully fetched content
  • •Report failures to user in summary

No Valid Content

  • •If all URL fetches fail, inform the user
  • •Ask for alternative URLs or direct content paste

Subagent Failures

  • •If any subagent fails, report the error
  • •Attempt to continue with available outputs
  • •Inform user which step failed and why

Important Notes

  • •Always use the same timestamp across all files in a single run for traceability
  • •Preserve all generated files—never overwrite previous runs
  • •Each subagent call should include complete context (they have no memory)
  • •Wait for user input during Phase 1 context gathering
  • •Generate exactly 3 variations in Phase 2