Doc Architecture Review

Documentation Architecture Review

Evaluate whether documentation is organized so that readers can find what they need, understand where they are, and navigate efficiently. The output is an architecture assessment with specific restructuring recommendations — not new content.

When to Use

• When restructuring or reorganizing documentation
• When adding a new section or doc type to an existing set
• When users report "I know it's documented somewhere but can't find it"
• When the doc set has grown organically and needs rationalization
• After doc-completeness-audit identifies gaps — before filling them, ensure the structure can accommodate new content
• Periodic review of navigation and discoverability

Quick Reference

Workflow Overview

Phase 0: Personas    → Establish the doc set's primary 1-3 personas
Phase 1: Map         → Build the current doc structure map (incl. link graph)
Phase 2: Evaluate    → Score against IA heuristics, parameterized by personas + doc type
Phase 3: Model       → Compare structure to user mental models per persona
Phase 4: Report      → Produce the architecture review with per-persona findings

Phase 0: Establish Personas

A "good" architecture is good for someone specific. Without personas, the heuristics apply a default standard that systematically misjudges docs serving non-default audiences (a flat reference doc scored as "poorly hierarchical" because it doesn't follow Quick Start → advanced).

Step 0a — Identify the doc set's audiences

Read the doc set's entry pages (README, index.md, landing pages) and the highest-traffic top-level docs. Identify which 1–3 personas from references/personas.md are the primary readers. Common patterns:

Step 0b — Draft persona profiles

For each identified persona, copy the profile from references/personas.md verbatim. Don't paraphrase — the explicit profile is what calibrates downstream sub-agents. If a persona almost fits but a dimension differs, define a custom persona using the same five-field structure.

Step 0c — Note conflicts

If the doc set serves more than one persona with conflicting needs (e.g., Onboarding User wants narrative, API Looker-Up wants terseness), note this explicitly. The synthesis report will surface where current structure favors one persona at the cost of another.

Output: A persona block (1–3 personas + any conflict notes) that feeds every downstream sub-agent prompt.

Phase 1: Map the Current Structure

Build a complete picture of the documentation architecture.

Step 1a: Physical Structure

Generate the file tree of all documentation:

find docs/ site/ -name '*.md' -o -name '*.html' | sort

Record:

• Directory hierarchy and nesting depth
• File count per directory
• Naming conventions (kebab-case, snake_case, mixed)

Step 1b: Navigation Structure

Identify every way a reader can navigate:

Step 1c: Entry Points

Identify how readers arrive:

• Direct — typing a URL or bookmarking
• Search — site search or external search engine
• Navigation — sidebar, breadcrumb, landing page links
• Cross-reference — link from another doc page
• External — README, GitHub, blog post, error message linking to docs

Map which pages are reachable from each entry point. Pages unreachable from common entry points are effectively invisible.

Step 1d: Link Graph (mechanical)

Run the bundled link graph analyzer to extract deterministic facts about inter-doc linking:

python3 skills/doc-architecture-review/scripts/link_graph.py --scope all --json > graph.json

# Or human-readable:
python3 skills/doc-architecture-review/scripts/link_graph.py --scope all

The script produces:

• Orphans — pages with no inbound links (excluding entry points like index.md and README.md). Direct input to Heuristic 1 (Findability).
• Dead-ends — pages with no outbound links. Content silos.
• Reciprocity ratio — fraction of edges that have a back-link. Direct input to Heuristic 4 (Cross-Linking Quality).
• Hubs — pages with high in-degree. Natural reference targets.
• Broken links — internal links that don't resolve. Direct input to Heuristic 4.

These are mechanical facts, not judgments. The judgment-heavy parts of Heuristics 1 and 4 (are links contextual? do navigation labels use user language?) are evaluated by sonnet sub-agents in Phase 2.

Output: A structure map with physical hierarchy, navigation paths, entry points, and the link graph JSON.

Phase 2: Evaluate Against IA Heuristics

Assess the structure against seven heuristics. Load references/ia-heuristics.md for detailed scoring criteria.

Mechanical vs judgment split

For a doc set of any meaningful size, the orchestrator can't read every page to score every heuristic — that strains the context window and produces patchy evaluation. Phase 2 splits work:

• Mechanical part — driven by the Phase 1d link graph JSON. Orphan counts, reciprocity ratio, broken-link counts, hub identification: these are facts, not judgments. The orchestrator reads the JSON and assigns scores deterministically.
• Judgment part — dispatched to general-purpose + sonnet sub-agents organized by heuristic. Each agent receives a focused slice of the doc set and returns specific findings with citations.

Sonnet sub-agent dispatch

Three judgment-heavy heuristics warrant dedicated agents. Each agent's prompt inlines the persona block from Phase 0 and the relevant doc-type criteria from references/ia-heuristics.md. The agent scores per-persona, not against a generic default.

Agent 1 — Findability narrative review (Heuristic 1):

subagent_type: "general-purpose"
model: "sonnet"
description: "Findability narrative review"

Prompt template:

Read landing pages, navigation configs (_config.yml, front-matter
nav_order/parent), and the orphans list from the link graph JSON.

Personas (from Phase 0):
<INLINE PERSONA BLOCK — full profile per persona, not summary>

Doc-type criteria for Heuristic 1:
<INLINE Heuristic 1 section from references/ia-heuristics.md>

For each persona, score Findability 1-5 and identify specific failures:
- Are navigation labels in this persona's language?
- Are entry points appropriate for how this persona arrives?
- Are orphans concentrated in a doc type that fails this persona's task?

Output per-persona scores plus findings. When personas conflict (e.g.,
nav labels in one's language fail another), surface the conflict
explicitly rather than averaging.

Agent 2 — Cross-linking quality review (Heuristic 4):

subagent_type: "general-purpose"
model: "sonnet"
description: "Cross-link quality review"

Prompt template:

Read 5-10 representative pages across doc types. Read the link graph
JSON's reciprocity statistics.

Personas (from Phase 0):
<INLINE PERSONA BLOCK>

Doc-type criteria for Heuristic 4:
<INLINE Heuristic 4 section from references/ia-heuristics.md — note
the per-doc-type linking patterns table>

For each persona, score Cross-Linking 1-5. Distinguish:
- Are links contextual (explain why to follow)? Low priority for
  Looker-Up reference scanning, high priority for Onboarding User
  exploration.
- Does linking density match the doc type's pattern?
- Are there mutual links between related concepts (high reciprocity)
  where appropriate?

Agent 3 — Pattern consistency review (Heuristic 5, per-doc-type dispatch):

For each doc type present (reference, tutorial, guide, explanation, ADR, runbook, README), dispatch one sonnet agent with all docs of that type:

subagent_type: "general-purpose"
model: "sonnet"
description: "Consistency review for <doc-type>"

Prompt template:

Examine all <DOC_TYPE> docs in the set: <LIST_OF_PATHS>.

Persona affected (from Phase 0):
<INLINE PERSONA BLOCK for this doc type's primary persona>

Expected template for <DOC_TYPE> per references/ia-heuristics.md
Heuristic 5:
<INLINE template signals row>

Identify the implicit template — common section headings, structural
conventions — and flag pages that deviate. For each deviation:
- Is it intentional (handles a special case the template doesn't
  cover)? Note the reason.
- Is it accidental (older page; different author; conventions hadn't
  settled)? Flag for harmonization.

Score 1-5 for within-type consistency.

Heuristic scoring

Heuristic 1: Findability

Can readers locate information without knowing where it lives?

Check:

• Orphaned pages (no inbound links, not in navigation)
• Dead ends (pages with no outbound links to related content)
• Search coverage (are all pages indexed? do headings use searchable terms?)
• Navigation labels (do they use user language or developer jargon?)

Heuristic 2: Hierarchy Coherence

Does the nesting make sense? Can a reader predict where to find something?

Check:

• Depth — flag anything nested >3 levels
• Breadth — flag directories with >10 immediate children (consider subcategories)
• Sibling coherence — are items at the same level truly peers?
• Naming — do directory names describe contents from the reader's perspective?

Heuristic 3: Progressive Disclosure

Does the doc set layer information from simple to complex?

Check:

• Quick start exists and is prominently linked
• Getting started path is linear and completable in <15 minutes
• Advanced topics are separated from basics, not interleaved
• Each doc states its prerequisites
• Cross-references point readers to deeper material, not shallower

Heuristic 4: Cross-Linking Quality

Do links between pages create useful connections or noise?

Check:

• Link density — pages with zero outbound links, pages with >20
• Reciprocity — if A links to B, does B link back (where appropriate)?
• Context — links explain why the reader would follow them, not just "click here"
• Anchor precision — links go to the right section, not just the right page
• Broken links — links that resolve to 404 or wrong content

Heuristic 5: Consistency of Patterns

Do similar pages follow similar structures?

Check:

• Do all reference pages have the same sections?
• Do all tutorials follow the same progression?
• Do all guides have prerequisites and next steps?
• Are metadata conventions (front matter, titles, descriptions) consistent?

Heuristic 6: Separation of Concerns

Are different doc types (reference, tutorial, guide, explanation) kept distinct?

Check:

• Pages that mix "how to" with "what it is" with "API details"
• Tutorials that double as reference (readers can't scan for a specific flag)
• Reference pages that include narrative explanations better suited to guides
• Use the Diataxis framework as a lens: tutorials, how-to guides, reference, explanation

Heuristic 7: Maintenance Burden

Is the structure sustainable as docs grow?

Check:

• Is there a clear directory/category for new feature docs?
• Are naming conventions documented and followed?
• Would doubling the doc set break the navigation?
• Are there catch-all directories growing without bounds?

Phase 3: Mental Model Comparison

Compare the documentation structure to how users actually think about the product.

User Mental Models

Identify the primary mental models users bring:

Most doc sets serve multiple models. The question is: which model does the navigation reflect, and does it match the primary user need?

Mismatch Indicators

• Users search for task-based terms but docs are organized by feature
• Getting started guide assumes feature knowledge the reader doesn't have yet
• Navigation uses internal terminology that users don't recognize
• Users land on the right page but can't find the right section

Phase 4: Produce the Architecture Review

Report Format

# Documentation Architecture Review

**Review date:** YYYY-MM-DD
**Scope:** [doc set reviewed]
**Total pages:** N
**Max depth:** N levels
**Orphaned pages:** N
**Personas evaluated:** [comma-separated list from Phase 0]

---

## Personas

[Inline the Phase 0 persona block — full profile per persona, plus any conflict notes]

---

## Summary

[2-3 sentences: overall architecture assessment, per-persona where relevant]

Heuristic scores per persona:

| Heuristic | Persona A | Persona B | Persona C | Notes |
|-----------|-----------|-----------|-----------|-------|
| Findability | N/5 | N/5 | N/5 | [one line, surface persona conflicts] |
| Hierarchy Coherence | N/5 | N/5 | N/5 | [one line] |
| Progressive Disclosure | N/5 or N/A | N/5 or N/A | N/5 or N/A | [N/A is valid for reference/ADR — see rubric] |
| Cross-Linking Quality | N/5 | N/5 | N/5 | [one line] |
| Consistency of Patterns | N/5 | N/5 | N/5 | [one line] |
| Separation of Concerns | N/5 | N/5 | N/5 | [one line] |
| Maintenance Burden | N/5 | N/5 | N/5 | [one line] |
| **Per-persona total** | **N/35** | **N/35** | **N/35** | |

Architecture grade per persona: [A / B / C / D / F]

When grades differ across personas, that's a finding, not a defect to
average away. Flag the structural bias toward whichever persona scores
highest.

---

## Structure Map

[File tree with annotations: orphan markers, depth warnings, misplacement flags]

---

## Critical Findings

### [Finding title]
**Heuristic:** [which]
**Impact:** [who is affected and how]
**Evidence:** [specific examples — pages, paths, search queries]
**Recommendation:** [specific restructuring action]

---

## Navigation Path Analysis

### Path: New User Onboarding
**Entry point:** [where they start]
**Goal:** [what they need to accomplish]
**Actual path:** [pages they traverse]
**Friction points:** [where they get lost or stuck]
**Ideal path:** [what it should be]

### Path: [Another key user journey]
...

---

## Orphaned Pages

| Page | Why It's Orphaned | Recommendation |
|------|-------------------|----------------|
| [path] | [no inbound links / not in nav] | [add to nav / link from X / archive] |

---

## Mental Model Alignment

**Primary user model:** [task / feature / role / chronological]
**Current structure model:** [which model the nav reflects]
**Alignment:** [match / partial / mismatch]
**Recommendation:** [restructure, add alternative navigation, or accept the gap]

---

## Restructuring Recommendations

Ordered by impact:

1. [Highest impact structural change]
2. [Second highest]
3. ...

---

## Strengths

[What's working well in the current architecture]

Integration with Other Doc Skills

doc-maintenance         →  Structural health (links, orphans, folders)
doc-claim-validator     →  Semantic accuracy (do claims match code?)
doc-completeness-audit  →  Topic coverage (is everything documented?)
doc-quality-review      →  Prose quality (is it well-written?)
doc-architecture-review →  Information architecture (is it findable?)

Run this skill after doc-completeness-audit — you need to know what's missing before evaluating whether the structure can accommodate it. Run before filling gaps, so new content lands in the right place.

Anti-Patterns

• Do not reorganize during the review — produce findings, not a new file tree
• Do not treat your preferred structure as the "correct" one — evaluate against user needs
• Do not evaluate archived docs (docs/archive/) — they are historical
• Do not confuse "I know where things are" with "a new user would know" — test with fresh eyes
• Do not recommend restructuring for its own sake — the cost of moving docs (broken links, muscle memory, bookmarks) must be justified by the improvement
• Do not ignore the cost of change — a mediocre-but-stable structure may be better than a perfect structure that requires moving 50 pages

Bundled Resources

References

• references/ia-heuristics.md — Detailed scoring criteria and examples for each heuristic