tachyon-beep/using-wiki-manager

Using Wiki Manager

Overview

This meta-skill routes you to the right document set management skills based on your task. Load this skill when you need to create, organize, maintain, or audit a collection of interrelated documents.

Core Principle: Managing a document set is a different discipline from writing a document. Individual document quality is necessary but not sufficient — the set must be consistent, navigable, self-sufficient at each depth tier, and maintainable over time.

When to Use

Load this skill when:

• Managing multiple related documents as a collection (standards suites, policy frameworks, technical documentation sets)
• User mentions: "wiki", "document set", "reading paths", "cross-reference", "consistency audit", "derivative", "propagate changes"
• Writing derivative documents from root/canonical sources
• Checking if documents are self-sufficient for their audience
• Propagating changes across a document collection
• Onboarding an existing collection of documents that lacks formal structure

Don't use for: Writing a single standalone document (use muna-technical-writer instead). Panel-testing documents with simulated readers (use muna-panel-review instead).

---

How to Access Reference Sheets

IMPORTANT: All reference sheets are located in the SAME DIRECTORY as this SKILL.md file.

When this skill is loaded from: skills/using-wiki-manager/SKILL.md

Reference sheets like content-derivation.md are at: skills/using-wiki-manager/content-derivation.md

NOT at: skills/content-derivation.md ← WRONG PATH

When you see a link like [content-derivation.md](content-derivation.md), read the file from the same directory as this SKILL.md.

---

Routing by Entry Pattern

Pattern 0: Onboarding an Existing Document Set

Symptoms: "I have documents but no structure", "How do I organize these?", "Bootstrap", "Get started with existing docs", "We have a standards suite but no manifest"

Use command: /onboard-docset if available, or follow this sequence manually.

Route to: Skills in order (matches /onboard-docset step sequence):

1. document-set-architecture.md — inventory documents, classify root/derivative, discover derivation lineage, assign depth tiers, build manifest
2. cross-document-consistency.md — bootstrap minimum viable registries (15-30 terminology entries, 10-20 claims) by harvesting from the existing documents. This step is load-bearing — skipping it leaves the audit phase with no objective standard to check against
3. reading-path-design.md — discover implicit personas and reading paths, build coverage matrix
4. document-governance.md — assign ownership, define review triggers, plan the baseline audit

The onboarding workflow builds registries from the documents rather than requiring them upfront. Start with minimum viable versions and grow through use.

---

Pattern 1: Creating a New Document Set

Symptoms: "Starting from scratch", "New documentation suite", "Greenfield", "Planning a multi-document standards effort"

Route to: Skills in sequential order:

1. document-set-architecture.md — define structure, manifest, depth tiers
2. reading-path-design.md — define personas and reading paths
3. content-derivation.md — set up derivation recipes for each derivative section
4. cross-document-consistency.md — establish terminology and claim registries
5. document-governance.md — assign ownership, define review triggers

document-evolution.md activates after the first version of the set is published.

---

Pattern 2: Day-to-Day Maintenance

Symptoms: Specific maintenance tasks on an existing, structured document set.

| User intent | Route to | |---|---| | Add or restructure a document | document-set-architecture.md | | Add a persona, check coverage, fix a dead-end path | reading-path-design.md | | Write or update a derivative document | content-derivation.md | | Fix terminology, verify claims, check links | cross-document-consistency.md | | Respond to upstream change (root document or external standard) | document-evolution.md | | Set up review workflows, assign ownership, run quality gates | document-governance.md |

---

Pattern 3: Change Propagation

Symptoms: "I updated the paper", "Root document changed", "Propagate changes", "Section X of the spec was rewritten"

Use command: /propagate-change if available, or chain manually:

1. Evolution (document-evolution.md) — classify the change, run impact trace → output: work list of affected derivative sections
2. Derivation (content-derivation.md) — for each affected section, re-apply the derivation recipe → output: updated derivative content
3. Consistency (cross-document-consistency.md) — verify claims and terminology match registries → output: defect list
4. Governance (document-governance.md) — surface quality gate results to human for review

This is the workhorse workflow — more targeted than a full audit, avoids overkill.

---

Pattern 4: Full Audit

Symptoms: "Check everything", "Audit", "Health check", "Pre-publish review", "Quarterly review"

Use command: /audit-docset if available, or run in sequence:

1. Consistency audit (cross-document-consistency.md) — terminology, claims, structural conventions, links
2. Derivation integrity (content-derivation.md) — recipe compliance, self-sufficiency, deferral audit
3. Path completeness (reading-path-design.md) — coverage matrix, stepping stone validation
4. Link integrity (cross-document-consistency.md) — forward links, bidirectional awareness, orphan sections, anchor stability

Output: unified defect list, triaged by priority (P1 wrong info → P2 degraded navigation → P3 friction).

---

Cross-Faction Integration

With muna-technical-writer

Use both when you need set-level structure AND individual document quality:

• Wiki Manager handles: manifest, derivation recipes, consistency registries, reading paths, governance
• Technical Writer handles: document structure (ADRs, APIs, runbooks), clarity and style, diagram conventions

Example: "Write a new CISO Assessment for our standards suite" → Load wiki-manager for the derivation recipe and self-sufficiency contract, load technical-writer for the document's internal structure and clarity.

With muna-panel-review

Wiki Manager ensures the set is structurally sound before testing. Panel Review tests how documents land with audiences after structure is in place.

The wiki-manager Persona Registry can feed panel-review's persona definitions: export the persona's task, depth tolerance, and domain vocabulary as the scenario framing for a panel-review session.

---

Reference Sheet Catalog

Core Skills

| Skill | Covers | |---|---| | document-set-architecture.md | Manifest, derivation graph, depth tiers, multi-root reconciliation, conflict resolution | | reading-path-design.md | Persona registry, path completeness, coverage matrix, time-based paths, stepping stone validation | | content-derivation.md | Derivation modes, recipes, self-sufficiency test, deferral audit — the anti-laziness discipline | | cross-document-consistency.md | Terminology/claim registries, granularity heuristics, link integrity, consistency audit, triage | | document-evolution.md | Change classification, impact traces, git integration, external dependencies, deprecation | | document-governance.md | Ownership model, LLM-as-steward trust model, quality gates, review triggers |

---

Design Principles

1. Self-sufficiency over linking — every document must stand alone for its audience
2. Explicit over implicit — derivation lineage, terminology choices, and ownership are declared, not assumed
3. Auditable — every quality claim has a concrete, repeatable test
4. Two-root aware — handles multiple canonical sources, including conflict resolution when roots disagree
5. Anti-laziness by design — derivation recipes make faithful synthesis the path of least resistance for LLMs
6. Governance ties it together — the other five skills define what to check; governance defines who checks and when
7. Bootstrappable — adoptable incrementally on existing document sets via minimum viable registries
8. Registries stay lean — granularity heuristics prevent registries from becoming maintenance burdens
9. Human-in-the-loop for trust — when Claude is the steward, quality gate results are surfaced, not silently self-approved
10. Git-native — manifests, registries, and documents are versioned together; change classification leverages diffs

---

Common Mistakes

Loading All Skills for Every Task

Wrong: Load all 6 reference sheets for every document set task. Right: Use the routing table above. A derivative update needs content-derivation.md and maybe governance. A terminology fix needs cross-document-consistency.md only.

Skipping the Manifest

Wrong: Jump straight to writing derivatives without declaring the document set structure. Right: Start with document-set-architecture.md. The manifest is the foundation everything else references.

Treating Derivation as Summarization

Wrong: "Summarize the paper for executives" → write a shorter version. Right: "Derive an executive brief from the paper using distillation mode" → follow the derivation recipe, run the self-sufficiency test, audit deferrals. Derivation is a discipline, not a writing task.

Auditing Without Registries

Wrong: Run a consistency audit by reading all documents and checking for problems. Right: Build the terminology and claim registries first (even minimum viable versions), then audit against them. Without registries, you have no objective standard to audit against.

---

Decision Tree

Starting a document set task?
├─ Do the documents have a manifest?
│  ├─ No → Pattern 0: Onboard (document-set-architecture first)
│  └─ Yes
│     ├─ What are you doing?
│     │  ├─ Writing/updating a derivative → content-derivation
│     │  ├─ Root document changed → Pattern 3: Change propagation
│     │  ├─ Adding a persona/fixing paths → reading-path-design
│     │  ├─ Fixing terminology/claims/links → cross-document-consistency
│     │  ├─ Setting up ownership/review process → document-governance
│     │  └─ Full health check → Pattern 4: Full audit
│     └─ Creating a new document set? → Pattern 1: Sequential setup