llms-txt

Storytelling (WillWrite)

AI-powered narrative generation — transforming prompts into complete, multi-chapter stories through advancing patterns, grounded in Creative Orientation and Indigenous ceremonial technology.

---

What Is Storytelling?

Storytelling (also known as WillWrite or SpecForge) is a Python package that empowers users to manifest complete, coherent narratives from a single creative prompt. It orchestrates a multi-stage generation pipeline using LangGraph state machines, multiple LLM providers, knowledge integration (RAG), and persistent sessions that survive interruptions.

What distinguishes it from other text generation tools:

• Creative Orientation — built on Robert Fritz’s structural dynamics, not problem-solving
• Ceremonial Integration — Indigenous five-phase methodology via the IAIP bridge
• Session Persistence — PHOENIX_WEAVE checkpoint/resume across any interruption
• Knowledge-Aware — RAG from local files, web content, and CoAiAPy datasets
• LLM-Agnostic — URI-based provider system (Google, Ollama, OpenRouter, custom endpoints)

Resource	URL
llms.txt	storytelling.jgwill.com/llms.txt
llms-full.txt	storytelling.jgwill.com/llms-full.txt
Source	github.com/jgwill/storytelling
Specifications	rispecs/

---

Generation Pipeline

The story generation pipeline follows four Creative Advancement Scenarios:

Scenario	Phase	What Happens
1. Story Foundation	Prompt → Elements → Outline	Extract meta-instructions, generate characters/plot/theme, create chapter structure
2. Chapter Crafting	Chapter × Scene × Layer	4 scenes per chapter, 4 layers per scene (plot → character → dialogue → integration)
3. Polish & Finalize	Edit → Scrub → Translate → Metadata	Holistic consistency, artifact removal, optional translation, title/summary/tags
4. Knowledge Augmentation	RAG at outline + chapter stages	Inject relevant context from markdown files, web, and CoAiAPy datasets

Each stage is a checkpoint — sessions can be resumed from any point.

---

Architecture Diagrams

The README.md at storytelling.jgwill.com includes three Mermaid architecture diagrams:

1. Story Generation Pipeline — full LangGraph node graph with STC phases (Germination / Assimilation / Completion)
2. STC State Machine — pipeline stages as creative phase state transitions; reveals oscillation risk in the chapter revision loop
3. NarrativeAware Enrichment Loop — Three-Universe Analysis → Emotional Scoring → Gap Identification → Enrichment cycle

Plus the Wâpano (NARINTEL:EAST) envisioned architecture — the Four Directions multi-agent system that begins the structural decolonization of this package.

---

Component	Purpose
Story Graph (graph.py)	LangGraph state machine orchestrating the full pipeline
Session Manager (session_manager.py)	PHOENIX_WEAVE persistence with checkpoint/resume
LLM Providers (llm_providers.py)	URI-based multi-provider abstraction; different models per stage
RAG (rag.py)	Enhanced multi-source knowledge retrieval (FAISS + web + CoAiAPy)
IAIP Bridge (iaip_bridge.py)	Five-phase ceremonial methodology for story generation
MCP Server (storytelling_mcp/)	Model Context Protocol for LLM agent orchestration

Tiered Dependencies

Tier	Size	Includes
Core	~50MB	Full generation, LangGraph, LangChain, Pydantic
+RAG	+3-5GB	Sentence-transformers, FAISS, document processing
+Specialized	Variable	Provider SDKs, GPU, cloud integrations

---

Relationship to This Portfolio

Storytelling is a primary Ceremonial Technology application — a concrete manifestation of the frameworks documented across this portfolio:

Framework	How Storytelling Embodies It
Creative Orientation	All specifications use creation language; narratives are advanced into being
Structural Tension	Specs use Structural Tension Blocks; generation holds tension through delayed resolution
RISE Framework	rispecs/ are RISE-generated blueprints
Narrative Craft	The package generates the narratives that beats then document
Ceremonial Technology	IAIP bridge implements five ceremonial phases
Indigenous Research Paradigm	Two-Eyed Seeing balances Western AI with Indigenous wisdom
Performance Truth	Critic/revision loops embody the MMOT process

The IAIP bridge maps story generation to five Indigenous ceremonial phases:

1. Miigwechiwendam — Sacred space and intention setting
2. Nindokendaan — Two-eyed research and knowledge gathering
3. Ningwaab — Knowledge integration through relational synthesis
4. Nindoodam — Creative expression within ceremonial container
5. Migwech — Ceremonial closing, reflection, and wisdom capture

---

RISE Specifications

The rispecs/ directory contains complete implementation-agnostic specifications:

Spec	Focus
ApplicationLogic	Story generation pipeline and graph orchestration
Session_Management_Architecture	PHOENIX_WEAVE checkpoint/resume
RAG_Implementation_Specification	Knowledge base integration
LLM_Provider_Specification	URI-based multi-provider abstraction
IAIP_Integration_Specification	Indigenous ceremonial framework
Narrative_Intelligence_Integration	Character arcs, emotional beats, thematic continuity
Tiered_Package_Architecture	Progressive dependency tiers

All governed by the Creative Orientation Operating Guide and Terms of Agreement.

---

Key Sources

• llms.txt — compact index for LLM consumption
• llms-full.txt — expanded index with detailed summaries
• Source repository
• Creative Orientation — foundational framework
• RISE Framework — specification methodology
• Narrative Beats — story documentation
• Structural Tension Charts — advancement methodology

This site is open source. Improve this page.