wiki-doc-generator

name	wiki-doc-generator
description	Generate structured wiki documentation from brand campaign outputs and business artifacts using parallel agent dispatch.

Execution Context

This skill runs AFTER the brandmint pipeline completes, not as part of it. It reads JSON outputs from .brandmint/outputs/ and generates wiki markdown. Always use parallel agents (Task tool) for document generation — never generate pages sequentially. If running from bm launch, ensure all waves have completed first.

Wiki Documentation Generator

Transform campaign outputs and business documents into structured, interconnected wiki documentation using parallel agent dispatch for maximum efficiency.

Workflow Overview

Phase 1: Inventory        → Scan input documents, classify by type
Phase 2: Dispatch Agents  → Parallel processing of document categories
Phase 3: Cross-link       → Build navigation and internal references
Phase 4: Output           → Generate wiki-ready markdown structure

Phase 1: Inventory Source Documents

Scan the input location for source documents. Classify each by type:

Document Type	Source Skill	Wiki Section
MDS (Messaging Direction Summary)	mds-messaging-direction-summary	Product Overview, Features
Buyer Persona	buyer-persona	User Personas
Product Positioning	product-positioning-summary	Brand Positioning
Voice & Tone	voice-and-tone	Brand Guidelines
Competitor Analysis	competitor-analysis	Market Context
Detailed Product Description	detailed-product-description	Product Specs
Campaign Page Copy	campaign-page-copy	Marketing Assets
Email Sequences	welcome/pre-launch/launch-email-sequence	Communications
Proposal	thoughtseed-proposal-generator (planned)	Technical Architecture, Timeline
Contract	thoughtseed-contract-generator (planned)	Project Scope, Deliverables

Run inventory script to classify:

python3 scripts/inventory-sources.py /path/to/source/documents

Visual Asset Inventory

If the brand has generated visual assets (via bm visual), also generate the asset-to-wiki mapping:

python3 scripts/map-assets-to-wiki.py /path/to/brand/generated/

This produces wiki-asset-map.json which maps each visual asset to its target wiki page, section, and role (hero, inline, gallery, or meta). Pass this mapping to agents in Phase 2 so they embed images in the correct locations.

Phase 2: Dispatch Parallel Agents

Each document category becomes an independent agent task. Process 3-5 categories in parallel using the dispatching-parallel-agents pattern.

Agent Categories

Agent 1: Foundation Docs

• Sources: MDS, Product Positioning, Detailed Product Description
• Outputs: product/overview.md, product/features.md, product/specifications.md

Agent 2: Brand Docs

• Sources: Voice & Tone, Visual Identity (if present)
• Outputs: brand/voice-tone.md, brand/visual-guidelines.md, brand/writing-principles.md

Agent 3: Persona Docs

• Sources: Buyer Persona, Competitor Analysis
• Outputs: audience/primary-persona.md, audience/secondary-personas.md, market/competitive-landscape.md

Agent 4: Marketing Assets

• Sources: Campaign Copy, Email Sequences, Ad Copy
• Outputs: marketing/campaign-copy.md, marketing/email-templates.md, marketing/ad-creative.md

Agent 5: Project Docs (if Proposal/Contract present)

• Sources: Proposal, Contract
• Outputs: project/architecture.md, project/timeline.md, project/team-roles.md

Agent 6: Visual Assets (if wiki-asset-map.json exists)

• Sources: wiki-asset-map.json, generated image directory
• Outputs: brand/visual-assets.md — a gallery page listing ALL visual assets organized by category
• Schema: See references/doc-schemas.md → Visual Asset Library

Agent Task Template

## Task: Generate {Category} Wiki Documentation

**Input Sources:**
{List of source files with paths}

**Output Requirements:**
1. Extract key information following the schema in references/doc-schemas.md
2. Generate markdown files with proper frontmatter
3. Include cross-references to related documents (use relative paths)
4. Maintain heading hierarchy (H1 = page title, H2 = sections)
5. If `wiki-asset-map.json` is available, embed images for your pages using `![Alt](/images/filename.png)` syntax. Place hero images after the H1. Place inline images within their mapped sections.

**Output Location:** /home/claude/wiki-output/{category}/

**Quality Checklist:**
- [ ] Source content synthesized (no copy-paste dumps)
- [ ] Frontmatter includes: title, description, category, tags
- [ ] Internal links use relative paths
- [ ] Code/technical content in fenced blocks

Return: Summary of generated files and any content gaps found.

Phase 3: Cross-Link and Structure

After agents return, build navigation and cross-references.

Navigation Structure

Generate navigation.yaml:

- title: Getting Started
  items:
    - index.md
    - quickstart.md

- title: Product
  items:
    - product/overview.md
    - product/features.md
    - product/specifications.md

- title: Brand
  items:
    - brand/voice-tone.md
    - brand/visual-guidelines.md

- title: Audience
  items:
    - audience/primary-persona.md
    - audience/secondary-personas.md

- title: Marketing
  items:
    - marketing/campaign-copy.md
    - marketing/email-templates.md

- title: Project
  items:
    - project/architecture.md
    - project/timeline.md

Cross-Reference Rules

1. Persona → Product: Link user needs to features that address them
2. Voice → Marketing: Link tone guidelines to content examples
3. Architecture → Timeline: Link technical components to delivery phases
4. Positioning → Competitor: Link differentiators to competitive gaps

Phase 4: Output Structure

wiki-output/
├── index.md                    # Homepage with project overview
├── navigation.yaml             # Sidebar configuration
├── getting-started/
│   └── quickstart.md
├── product/
│   ├── overview.md
│   ├── features.md
│   └── specifications.md
├── brand/
│   ├── voice-tone.md
│   ├── visual-guidelines.md
│   └── visual-assets.md          # Gallery of all visual assets (Agent 6)
├── audience/
│   ├── primary-persona.md
│   └── secondary-personas.md
├── market/
│   └── competitive-landscape.md
├── marketing/
│   ├── campaign-copy.md
│   ├── email-templates.md
│   └── ad-creative.md
└── project/
    ├── architecture.md
    ├── timeline.md
    └── scope.md

Document Frontmatter Schema

Every generated markdown file includes:

---
title: "Page Title"
description: "One-line description for SEO and navigation"
category: "product|brand|audience|marketing|project"
tags: ["relevant", "keywords"]
sources: ["mds.md", "positioning.md"]
lastUpdated: "2025-01-24"
---

Scenario-Based Depth

Adapt output based on project context:

Scenario	Pages	Sections Included
Brand Genesis ($3K)	5-8	Foundation + Brand basics
Crowdfunding Lean	12-15	+ Personas + Marketing basics
Crowdfunding Full	20-25	+ Full marketing + Project
DTC Launch	15-18	All sections, moderate depth
Enterprise GTM	25-30	All sections, maximum depth

Integration with Astro

After generating wiki docs, build with markdown-to-astro-wiki:

# Wiki output location
/home/claude/wiki-output/

# Initialize Astro project
./scripts/init-astro-wiki.sh project-wiki

# Process markdown WITH images (if generated/ directory exists)
./scripts/process-markdown.sh /home/claude/wiki-output ./project-wiki/src/content/docs --images /path/to/brand/generated/

# Build
cd project-wiki && bun run build

The --images flag copies all visual assets (PNG, JPG, WebP) from the generated directory into public/images/ so that ![Alt](/images/filename.png) references in markdown resolve correctly.

Error Handling

• Missing sources: Generate placeholder with TODO markers
• Conflicting info: Flag in output, cite both sources
• Incomplete sources: Extract available content, note gaps

Resources

• references/doc-schemas.md - Detailed schemas for each wiki page type
• references/frontmatter-templates.md - Frontmatter examples by category
• scripts/inventory-sources.py - Scan and classify input documents (text + visual)
• scripts/map-assets-to-wiki.py - Map visual assets to wiki pages
• scripts/validate-wiki.py - Check output structure and links