Skip to content
Wholesome Story Pipeline Skill logo

Wholesome Story Pipeline Skill

Generate full children's novels from seed files using the proven Creek pipeline approach.

mmcmedia/openclaw-agents0installs2stars

SKILL.md

Full skill instructions

Wholesome Story Pipeline Skill

Generate full children's novels from seed files using the proven Creek pipeline approach.

Overview

This is the Creek pipeline — the approach that produced the best stories (Library Cat, Puddle) at the lowest cost. The key insight: trust the AI to tell a story naturally, then polish once at the end. No quality loops during generation, no critique passes, no complex feedback cycles.

Pipeline Phases

PhaseModelPurposeCost/​Chapter
1. EXPANDGPT-5.2Seed → 4,500-5,000 words full prose~$0.07
2. EDITGPT-5-miniOne polish pass (90-100% retention)~$0.01
3. ASSEMBLELocalStitch edited chapters → clean markdownFree

Total cost: ~$0.08/​chapter, ~$0.50 for a 6-chapter story, ~$1.80 for a 23-chapter novel.

Prerequisites

  • OpenAI API key in /​projects/​wholesome-library-2026/​generation/​.env (OPENAI_API_KEY)
  • Seed file with ## Chapter headers (each chapter = 100-1500 word outline)
  • Python 3 available

Usage

Quick Run (Single Story)

cd /​Users/​mmcassistant/​clawd/​projects/​wholesome-library-2026/​generation
nohup python3 -u run-story.py <seed-file> <output-prefix> > /​tmp/<name>-expansion.log 2>&1 &

IMPORTANT: Always use nohup to prevent process death from exec timeouts.

Examples

# Novel-length (23 chapters, ~2 hours)
nohup python3 -u run-story.py seeds/​seed-frequency-full.md frequency > /​tmp/​frequency.log 2>&1 &

# Short story (6 chapters, ~30 min)
nohup python3 -u run-story.py fresh-seeds/​seed-02.md clockwork-sparrow > /​tmp/​clockwork.log 2>&1 &

Monitor Progress

tail -f /​tmp/<name>-expansion.log
# Or check output files:
ls -lh <output-prefix>-ch*-expanded.md  # Phase 1 outputs
ls -lh <output-prefix>-ch*-edited.md    # Phase 2 outputs
ls -lh <output-prefix>-final-clean.md   # Phase 3 output

Check Completion

# Word count of final output
wc -w <output-prefix>-final-clean.md

# AI tell count (lower = better, target <20 for full novel)
grep -ciE 'as if|as though|seemed to|felt a |couldn.t help but|found herself|found himself|washed over|settled over' <output-prefix>-final-clean.md

Seed File Format

Seeds must have this structure:

# Story Title

**Characters/​metadata at top (optional)**

## Chapter 1: Title
Outline text (100-1500 words describing what happens)

## Chapter 2: Title  
Outline text...

## Chapter 3: Title
Outline text...

The script parses ## Chapter headers. Everything before the first chapter header becomes the story context/​header passed to GPT-5.2.

Seed Sources

LocationDescription
generation/​seeds/Novel-length seeds (The Frequency, etc.)
generation/​fresh-seeds/Short story seeds (5 stories, 6 chapters each)
generation/​batch1/Batch 1 seeds (Puddle, Window, Birds, Chalk, Tree)

Pipeline Script Details

Script: generation/​run-story.py

Phase 1: EXPAND (GPT-5.2)

  • Each seed chapter → ~4,500-5,000 words of full novel prose
  • Feeds previous chapter ending for continuity
  • Rules baked into prompt: distinct character voices, show-don't-tell, varied pacing, concrete stakes
  • Max tokens: 16,000 per chapter
  • Skip logic: won't re-expand if output file already exists with >500 bytes
  • Automatic retry on short outputs (<50 words)

Phase 2: EDIT (GPT-5-mini)

  • One pass: structural + prose polish
  • Aggressive AI tell removal (as if, seemed to, felt a, etc.)
  • 90-100% word retention target — polish, NOT rewrite
  • Fixes pacing, echo words, weak verbs, body language variety
  • Skip logic: won't re-edit if edited file already exists

Phase 3: ASSEMBLE

  • Stitches all edited chapters (falls back to expanded if no edit)
  • Strips changelog artifacts
  • Cleans excessive newlines
  • Outputs: <prefix>-final-clean.md
  • Reports: final word count, AI tell count, total cost

Key Rules (DO NOT CHANGE)

  1. Use run-story.py — NOT run-story-option4.py (Option 4 adds complexity without quality improvement)
  2. GPT-5.2 for expansion, GPT-5-mini for editing (never GPT-4o, it's deprecated)
  3. Always use nohup for background execution
  4. Trust then polish — no quality loops during generation
  5. One edit pass is enough — multiple micro-edits make things worse

Post-Pipeline Steps

After the pipeline produces <prefix>-final-clean.md:

  1. Polish Pipeline (optional): cd polish-pipeline && ./​batch-polish.sh — triple helix editor for final quality pass
  2. Cover Generation: scripts/​generate-cover.sh — Nano Banana cover art
  3. Upload to Supabase: scripts/​upload-story.mjs — publish to Wholesome Library
  4. McKinzie Review: Move to review-queue/​pending/ for approval

Troubleshooting

IssueFix
Process dies mid-runRestart with same args — skip logic resumes from last completed chapter
Short chapter output (<200w)Script auto-retries once; if still short, check API quota
High AI tell count in finalRun polish pipeline for aggressive cleanup
Cost seems highCheck if GPT-5.2 is using reasoning tokens (reasoning:0 = good)

History

  • Feb 9, 2026: Creek pipeline identified as best approach (Library Cat test)
  • Feb 10, 2026: Batch 1 generated (5 stories: Puddle, Window, Birds, Chalk, Tree)
  • Feb 11, 2026: Skill formalized. "The Frequency" + "Clockwork Sparrow" running.

More skills from mmcmedia

Tempest Mockups Skill logo
mmcmedia/openclaw-agents

Tempest Mockups Skill

Generate Etsy product mockups automatically using the Tempest Photoshop script.

2 0
View
Analytics & Insights Expert logo
mmcmedia/openclaw-agents

Analytics & Insights Expert

Comprehensive data analysis specialist covering basic metrics through advanced predictive analytics and marketing ROI. Your go-to expert for all data-driven decision making.

2 0
View
Competitive Intelligence logo
mmcmedia/openclaw-agents

Competitive Intelligence

Competitor research and market analysis specialist who tracks what your competitors are doing and identifies opportunities they're missing.

2 0
View
Test-Driven Skill Creation logo
mmcmedia/openclaw-agents

Test-Driven Skill Creation

Create and validate skills by measuring the gap between model output WITH and WITHOUT the skill.

2 0
View
react-best-practices logo
mmcmedia/openclaw-agents

react-best-practices

React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimizati...

2 0
View
playwright-cli logo
mmcmedia/openclaw-agents

playwright-cli

Browser automation via Playwright CLI. Open pages, interact with elements, take screenshots, and more. Ideal for coding agents and automated testing workflows.

2 0
View
ux-audit logo
mmcmedia/openclaw-agents

ux-audit

AI skill for automated design audits. Evaluate interfaces against proven UX principles for visual hierarchy, accessibility, cognitive load, navigation, and more. Based on Making UX Decisions by Tommy Geoco.

2 0
View
Etsy Image Prompts Skill logo
mmcmedia/openclaw-agents

Etsy Image Prompts Skill

Generate high-quality, print-ready digital art for Etsy listings using AI image generators. This skill provides prompt formulas, templates, and best practices for creating sellable wall art, printables, and digital products.

2 0
View
frontend-design logo
mmcmedia/openclaw-agents

frontend-design

Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.

2 0
View
ui-audit logo
mmcmedia/openclaw-agents

ui-audit

AI skill for automated UI audits. Evaluate interfaces against proven UX principles for visual hierarchy, accessibility, cognitive load, navigation, and more. Based on Making UX Decisions by Tommy Geoco.

2 0
View

Popular AI tools

Kaiber logo
Video

Kaiber

Generate, edit, and beat-sync AI video with leading models in one workspace.

Paid
View
Vimcal logo
Productivity

Vimcal

The world's fastest calendar for remote work

Free
View

Transform Your Design with AI Designer by ImgCreator.ai

Freemium
View
Akool AI logo
Content & writing

Akool AI

Revolutionizing Video Production with AI-Powered Creativity

Paid
View

Extend an image past the frame and let AI fill the new aspect ratio.

Freemium
View
StarByFace logo
Security

StarByFace

Discover your celebrity doppelgänger with StarByFace!

Free
View
C

ChainClarity explains 700+ crypto whitepapers in plain English, with layered summaries, comparisons, research tools, alerts, and a $4.99 Pro plan.

Freemium
View
Opus Clip logo
Coding & apps

Opus Clip

Opus.ai: Revolutionize Your Web Experience

Free
View