Miadi Storyweaver Orchestration Kit
RISE specifications for a Copilot-first storytelling orchestration kit that coordinates story creation through specialized agents, repeatable skills, review gates, and exportable story artefacts.
The future kit should live at copilot/miadi-storyweaver-orchestration-kit/. These rispecs are the source of truth for the implementation agent that will create that kit.
Mission issue: jgwill/miadi-orchestration-kit#9.
Structural Tension
- Desired Outcome: A reusable Miadi-native Copilot kit that can turn a creative prompt into a coherent story workspace through coordinated writing, research, critique, editing, continuity, and export agents.
- Current Reality: The storytelling package already demonstrates RISE-aligned narrative generation patterns, but this repository does not yet define a reusable orchestration kit that can run without importing or depending on that package.
- Natural Progression: Promote the durable storytelling patterns into implementation-agnostic Copilot orchestration specs, then export those specs into agents, skills, launch instructions, and review contracts.
Read Order
| File |
Purpose |
01-reverse-engineer.md |
Extracts the reusable patterns from RISE, llms-txt, the storytelling rispecs, and existing orchestration kits. |
02-intent.md |
Defines the kit vision, boundaries, non-goals, success criteria, and structural tension. |
03-pipeline-spec.md |
Specifies the end-to-end storytelling workflow from brief intake to final export. |
04-agent-roster.md |
Defines the agents needed to write, review, revise, and steward a story. |
05-skill-surface.md |
Defines the Copilot skills the implementation agent should create. |
06-state-and-handoff.md |
Specifies session state, artefact paths, handoff contracts, and resume semantics. |
07-copilot-plugin-export.md |
Defines the plugin folder, manifest, launch commands, and smoke tests. |
08-quality-gates-and-review.md |
Defines review gates, critique formats, consent gates, and final acceptance. |
09-source-ledger.md |
Records source provenance and the path mismatch discovered during this mission. |
10-session-episode-storyforms.md |
Extends Storyweaver to extract episodes, StoryForms, StoryBeats, and Story Settings from meaningful sessions. |
Implementation Target
The implementation agent should create a Copilot plugin with this expected shape:
copilot/miadi-storyweaver-orchestration-kit/
.github/plugin/plugin.json
README.md
agents/
storyweaver-orchestration-architect.md
creative-brief-archaeologist.md
world-and-character-architect.md
narrative-research-weaver.md
outline-architect.md
scene-writer.md
continuity-keeper.md
developmental-editor.md
critique-reviewer.md
revision-weaver.md
line-editor.md
cultural-protocol-steward.md
export-steward.md
skills/
storyweaver-session-bootstrap/SKILL.md
storyweaver-brief-intake/SKILL.md
storyweaver-story-bible/SKILL.md
storyweaver-outline-and-beats/SKILL.md
storyweaver-chapter-wave/SKILL.md
storyweaver-review-loop/SKILL.md
storyweaver-continuity-ledger/SKILL.md
storyweaver-cultural-protocol-check/SKILL.md
storyweaver-export-packet/SKILL.md
Operating Boundary
The implementation must not import, shell into, or require the existing jgwill/storytelling package. That package is source archaeology only. The Copilot kit should be prompt, agent, and artefact orchestration that can operate in any writing workspace.
Creative Advancement Scenarios
Scenario 1: Creative Brief To Story Bible
- Desired Outcome: The writer has a precise story bible that captures vision, constraints, audience, genre, themes, characters, world, and boundaries.
- Current Reality: A prompt may be vivid but incomplete, contradictory, or mixed with operational instructions.
- Natural Progression: The brief archaeologist extracts meta-instructions, the world-and-character architect manifests a story bible, and the continuity keeper prepares the first ledger.
- Achieved Outcome: The drafting agents can write without repeatedly rediscovering the story’s foundation.
Scenario 2: Story Bible To Reviewed Outline
- Desired Outcome: The writer has a chapter or beat outline with visible structural tension, character movement, stakes, and ending trajectory.
- Current Reality: Story foundations alone do not tell drafting agents where each scene should advance.
- Natural Progression: The outline architect creates the narrative map, the developmental editor evaluates structure, and the revision weaver applies selected improvements.
- Achieved Outcome: Chapters can be drafted against a coherent skeleton.
Scenario 3: Reviewed Outline To Drafted Chapters
- Desired Outcome: Each chapter advances plot, character, emotion, and theme while honoring the story bible.
- Current Reality: Isolated chapter drafting can drift in voice, continuity, pacing, or promises to the reader.
- Natural Progression: The scene writer drafts in layers, the continuity keeper records changes, and critique agents route observations into revision moves.
- Achieved Outcome: The story accumulates as a coherent manuscript rather than a stack of disconnected generations.
Scenario 4: Drafted Chapters To Export Packet
- Desired Outcome: The writer receives a complete story export with metadata, review notes, continuity ledger, and implementation-readable provenance.
- Current Reality: Final text alone loses the reasoning, sources, constraints, and revision history that make the story resumable.
- Natural Progression: The line editor polishes prose, the export steward assembles deliverables, and the orchestration architect records session closure.
- Achieved Outcome: The story workspace can be read, resumed, reviewed, or adapted by future agents.
Scenario 5: Session To Episode Memory
- Desired Outcome: A meaningful agent/workspace session becomes an auditable episode with StoryForms, StoryBeats, Story Setting, related work, and follow-up commissions.
- Current Reality: Operational sessions can create real narrative movement, but their meaning often remains trapped in chat, memory notes, or command output.
- Natural Progression: The session-episode branch reads sources, extracts setting and beats, names reusable StoryForms, and exports an episode packet without fictionalizing source facts.
- Achieved Outcome: The Miadi world learns from the session as durable story-context memory.