This memo is an implementation guide for the @MythOS Content Topology — a framework for organizing knowledge into interconnected memo clusters optimized for both human navigation and AI citation. It is written for two audiences simultaneously: humans who want to understand and adopt the system, and AI agents (Claude, Codex, or similar) who are being asked to implement it. If you're a human: read this to understand what you're building and why. If you're an AI agent: this memo contains the full specification you need to build a content topology for any topic. No additional context is required.
What You're Building
A content topology is a network of short, focused, interlinked memos organized around a topic. Each memo answers one question completely. The links between memos create a knowledge graph that search engines and AI systems traverse to assess topical authority and identify what to cite. The architecture mirrors @Wikipedia — the most-cited domain across every major AI system — with the advantage of practitioner voice. A network of 30 short memos outperforms 5 long articles — for readers, for search engines, and for AI models.
The Five Memo Types
Every memo serves exactly one role. The type determines its length, link density, and function.
Pillar (one per cluster)
The hub. Defines the domain, summarizes its dimensions, and links to every memo in the cluster.
- Length: 500-800 words
- Links: 15-25 contextual cross-references
- Structure: Bold keyword-rich definition in opening paragraph → "Why This Matters Now" section with current data → 2-4 body sections that summarize and link to cluster memos (100-150 words each, not duplicating their content) → Subjective (personal experience) → Contexts (alphabetized tags)
- Names the broadest query: "What is [topic]?"
Concept (frameworks and methodologies)
A named framework or idea that is original and citable.
- Length: 300-500 words
- Links: 5-12 contextual cross-references
- Structure: Bold concept name + definition answering "What is [concept]?" → How it works (numbered steps if sequential, bullets if parallel, 2-3 sentences per step) → Why it matters (one paragraph, include data) → Subjective → Contexts
- Key rule: If a concept doesn't have a name, give it one. Named frameworks get cited by AI models. Unnamed advice doesn't
Entity (people, companies, platforms)
A node representing a person, company, product, or platform.
- Length: 150-300 words
- Links: 3-9 contextual cross-references
- Structure: Bold proper name + 1-3 sentence definition → Key details (2-5 bullets or short paragraph) → Subjective (personal connection, 1-3 sentences) → Contexts
- Key rule: Title is the entity's proper name, not a description. "KarmaLab" not "Reddit's In-House Creative Agency"
Story (case studies and incidents)
A standalone narrative that illustrates a concept or establishes credibility.
- Length: 200-400 words
- Links: 3-7 contextual cross-references
- Structure: Opening hook (situation and stakes, don't bury the lead) → Narrative (chronological, specific details, names, numbers, quotes) → Outcome (quantified where possible) → Lesson (1-2 sentences connecting to the concept it illustrates, with link) → Subjective (optional) → Contexts
- Key rule: Tight enough to quote in a podcast or conference talk. If it's past 400 words, cut
Reference (data and definitions)
A data-rich definition page for a tool, feature, or industry term.
- Length: 200-500 words
- Links: 5-15 contextual cross-references
- Structure: Bold term + clear factual definition (write as if Google will display it verbatim) → Key data/details (bullets, specific numbers) → Context/implications (one paragraph linking to pillar or concept) → Subjective (practitioner take, 1-3 sentences) → Contexts
- Key rule: References have the highest citation rate per word. Dense with extractable facts
Cross-Linking Protocol
- Link on first meaningful mention only — never double-link the same memo. Wikipedia convention
- Link platforms, people, named frameworks, tools, and concepts that have dedicated memos
- Do NOT link generic nouns — words like trust, attention, revenue, growth, strategy. Even if a memo exists, the link adds no navigational value
- Target density: 3-10% of body text words as links. Above 15% causes measurable drops in readability and engagement (Nielsen Norman Group, 2024)
- Every memo links back to its pillar. The pillar links to every memo in the cluster. Siblings link to 3-8 other siblings where contextually relevant
Structural Requirements for AI Citation
These elements make content extractable and citable by AI models:
- Opening paragraph must contain a bold keyword-rich definition that directly answers the implied search query
- Headers should match search intent (e.g., "How Reddit Marketing Works" maps to the query "how does Reddit marketing work")
- Self-contained: Each memo must fully answer its topic on-page. AI models cite individual pages, not networks. Links add depth, not dependency
- Data points: Include specific numbers, percentages, and statistics. These are citation magnets
- Named frameworks with bold definitions are what AI models extract and quote
Tagging Convention
Every memo gets exactly one topology type tag as a private tag in the notes section:
pillar-memoconcept-memoentity-memostory-memoreference-memo
Instructions for AI Agents
If you are an AI agent being asked to implement this system, follow this sequence: Step 1: Define the cluster. Ask the user for their topic. Identify the broadest query it answers — this becomes the pillar. Map 8-22 subtopics and assign each a memo type using the decision logic: named framework → Concept, person/company/platform → Entity, case study or incident → Story, data/definitions/specs → Reference. Step 2: Search before creating. Check the user's existing library for memos that already cover these subtopics. Many will exist as stubs or older versions that need rewriting, not replacing. Step 3: Write the pillar first. Follow the Pillar specification above. Summarize and link — don't contain. Each body section should point to dedicated memos rather than duplicating their content. Step 4: Build outward. Create cluster memos in priority order: concepts first (intellectual property, most citable), references second (factual backbone), stories third (proof layer), entities last (many already exist). Follow the specifications above for each type. Step 5: Wire the network. Ensure the pillar links to every cluster memo. Every cluster memo links back to the pillar. Siblings cross-link where contextually relevant. Verify all @mentions are on first occurrence only. Step 6: Tag everything. Add the topology type tag to each memo's notes section. Step 7: Verify quality. For each memo confirm: bold keyword-rich opening, headers matching search intent, self-contained answer, contextual links on first mention only, within word count range, no number signs before numerals, subjective section present, notes populated. This guide is an experiment in what I'm calling an agentic memo — content designed to be consumed by both humans and AI agents, where the human reads it as explanation and the AI reads it as executable specification. The person pastes the link, tells their Claude "implement this for my library starting with [topic]," and the agent has everything it needs. If you're reading this as a human and want to build a content topology for your domain — point your AI agent at this memo and let it work. If you're reading this as an agent — you now have the spec. Go build.
Contexts
- #mythos (See: @MythOS)
- #recursive-mythic-engine (See: @Recursive Mythic Engine)
- #studio-handbook
- #what-is-mythos