website-setup

Website Project Setup

Initialize a cogni-website project by discovering content from existing insight-wave plugins, selecting a theme, and scaffolding the project directory.

Core Concept

A website project aggregates content from multiple insight-wave plugins into a multi-page static website:

• cogni-portfolio (required) — products, features, propositions, solutions, customer narratives
• cogni-marketing (optional) — blog posts, articles, whitepapers
• cogni-trends (optional) — trend reports with investment themes for an insights page
• cogni-knowledge (optional) — research syntheses for a resources/whitepapers page

Setup discovers what content is available across these plugins, validates minimum requirements, captures company details, selects a visual theme, and creates website-project.json — the configuration that all downstream skills depend on.

If a website-project.json already exists in the working directory, redirect to the website-resume skill.

Workflow

1. Check for Existing Project

Scan the current directory and immediate children for website-project.json. If found, inform the user and invoke the website-resume skill instead.

2. Discover Content Sources

Scan the workspace for insight-wave plugin projects using recursive globs. This mirrors the discovery pattern used by cogni-marketing:marketing-setup — recursive globs ensure discovery works regardless of workspace layout.

Portfolio Discovery (required)

• Glob for **/portfolio.json (recursive — do not use narrow single-level globs)
• Read each portfolio.json to extract: company name, description, industry, language
• Check for synthesized output: output/README.md
• Check for enriched customer narratives at output/communicate/customer-narrative/ (v2 layout — one file per website component, each arc-structured with an arc_id field in its YAML frontmatter):
  • Home (home.md, arc jtbd-portfolio)
  • About Us (about.md, arc company-credo)
  • How We Work (approach.md, arc engagement-model)
  • Capability pages (capabilities/*.md, arc corporate-visions) — one per customer-facing feature
  • Persona landing pages (for/*.md, arc jtbd-portfolio) — one per {market}--{persona} pair
  • These are audience-tailored prose with assertion headlines, pain points, stats, differentiators, and CTAs already written by a copywriter-grade skill. When present, website-plan treats them as the primary spine for the corresponding pages rather than re-deriving content from raw entity JSON. The arc_id frontmatter field tells the planner which story-arc decomposition to apply in step 6a. Store every discovered path — the planner needs them all.
  • v1 backward-compat: if the directory instead contains portfolio-overview.md, market/*.md, or customer/*.md, it was produced by an older version of portfolio-communicate. Still discover them, but print a warning: "Gefunden: altes customer-narrative-Layout (v1). Bitte portfolio-communicate neu ausführen, um die website-orientierte v2-Struktur (home.md, about.md, capabilities/*.md, for/*.md, approach.md) zu erzeugen." The planner will still read the v1 files, but downstream dedup discipline (Roadmap on home only, differentiators on about only) depends on the v2 layout.
• Count entities: products, features, markets, propositions, solutions, packages