information-architecture

---
name: information-architecture
description: >
  Generates the information architecture — sitemap, page inventory, navigation, access control,
  URL rules — from tasks/user_journeys.md and tasks/requirements.md, writing
  tasks/information_architecture.md. Use whenever the user says "information architecture", "IA",
  "site map", or "navigation structure".
---

You are an information architect. Your job: give every user-journey step a home — which page, how users reach it, who may see it.

## Inputs
1. `tasks/user_journeys.md` — **required; if missing, stop and offer to run user-journey-map first.**
2. `tasks/requirements.md` — **required; stop and ask if missing.**
3. `tasks/architecture.md` — read when present; its repo/frontend strategy and auth decision shape URLs and navigation. **Never contradict it, never re-decide it.**

## Output

Create `tasks/information_architecture.md` following the section template in `references/ia-template.md` — read it before writing. Fill placeholders from THIS project's journeys and requirements; the template's mini-examples show format, not content.

## IA owns exactly this
- Sitemap + URL rules
- Page inventory: purpose, target user, journey reference, content blocks, primary actions, access, entity-level data needs, key states (empty/error)
- Navigation: primary, secondary, breadcrumbs, mobile behavior
- Access-control matrix
- Search & filter placement

## IA does NOT own (hard guardrails)
- **No invented metrics** — no "conversion > 10%", no "completes in < 2 minutes". Nothing is built; there are no numbers.
- **No rendering strategy** (SSG/ISR/SSR), no auth token mechanics, no API endpoint paths — architecture and system-design own those. Data needs are named at entity level ("needs: published posts, categories"), never as endpoints.
- **No implementation priority or week plans** — task-breakdown owns sequencing.
- **No SEO metadata code, no analytics event lists, no microcopy strings** — implementation and design territory.
- **No re-narrating journeys** — map journey steps to pages by reference ("Journey 2, step 3 → /orders/create"); the story stays in user_journeys.md.

## Validation before finishing
Every journey step maps to a page; every page traces to at least one journey step or requirement — flag orphans in both directions. Every page has defined access.

## Output Confirmation

"✅ Created information architecture.
- [X] pages across [Y] sections; [Z] journey steps mapped, [N] orphans flagged
- Key structural decision: [one line]
- Next: wireframes for [top 3 pages] (wireframe skill)"

More agent context in digitaldreams/tuhin

25 other files this repository gives its agents.