SEO content research: model articles for the Step-1 batch

1. obsidian-publish-alternative.md — "obsidian publish alternative", "obsidian publish self-hosted"

Models.

• Unmarkdown — Obsidian Publish Alternatives — vendor comparison done right (see TL;DR). 15-row matrix, "different kind of alternative" reframe, three soft CTAs.
• Simon Späti — Open-Source Obsidian Publish Alternatives — ~400 words, ranks on freshness ("last updated" since 2022), aggressive internal wikilinking, and the page is a published digital garden. This is exactly our dogfooding play; his vault is our closest structural precedent.
• Obsidian Forum thread — what searchers actually ask; mine it for FAQ phrasing.

Why they rank. Exact-match long-tail title, pain-first intro ("Publish is great but expensive"), freshness signals, and either a feature matrix (Unmarkdown) or a live-demo effect (ssp.sh).

Searcher intent. Obsidian user paying or about to pay $8–10/mo, wants cheaper/self-hosted/more control. Bottom-of-funnel, tool-shopping.

trip2g lead-with (all shipped). Self-hosted, MIT, $0 license; two-way sync so the vault stays local and edits flow both ways; custom Jet templates (Publish has CSS snippets only); paywalls/monetization; Telegram publishing; MCP endpoint on the same notes; multi-domain.

Where Obsidian Publish legitimately wins. Zero setup, official and maintained by the Obsidian team, polished graph view, no server to run or update, predictable hosting. Say so plainly: "if you never want to think about a server, pay the $8 and stop reading."

2. trip2g-vs-quartz.md — "quartz alternative", "obsidian digital garden self-hosted"

Models.

• Juhis — Building a digital garden with Obsidian and Quartz — honest user write-up: admits Quartz's heavy client-side JS and doc gaps. His friction list is our comparison's raw material.
• Simon Späti — Quartz: Publish Obsidian Vault — pro-Quartz reference point ("huge community, features merged daily"); cite it as the fair steelman.
• XDA — How I Turned My Obsidian Notes into a Website Using Quartz — big-publisher how-to; shows the "no coding needed" hook Quartz owns.

Why they rank. First-person experience with real friction (hamatti), topical authority + freshness (ssp.sh), domain authority + benefit-hook headline (XDA).

Searcher intent. Knows Quartz, either evaluating it or hitting its walls (rebuild loop, git-per-edit, JS weight, no dynamic anything).

trip2g lead-with. No build step: edit in Obsidian → live in ~a second via two-way sync (vs edit → build → commit → deploy); dynamic features a static site can't have — paywalls, forms, subscriptions, RSS per base, live editing; MCP so agents query the same garden; per-domain routing.

Where Quartz legitimately wins. Completely free hosting (GitHub Pages), no server to operate or secure, huge community and theme/plugin velocity, graph view, static-site speed and simplicity. Honest close: "static + free beats a server if your garden is read-only."

3. trip2g-vs-hugo.md — "publish obsidian vault hugo"

Models.

• Jacob Kaplan-Moss — Publishing an Obsidian vault with Hugo — high-authority author; the article is a candid catalog of the pain (wikilink conversion, _index.md workarounds, pre-commit hooks, "knitting together" tools) and even opens by admitting "the easy way is $100/yr on Obsidian Publish". The friction he documents is precisely the pipeline trip2g deletes.
• Sagar Behere — Publishing Obsidian vault with Hugo and obsidian-to-hugo — the converter-script genre; shows how much custom glue the workflow demands.

Why they rank. Author authority (jacobian), exact-match workflow query, real code and admitted tradeoffs.

Searcher intent. Technical user mid-struggle: wants Obsidian authoring + a website, is fighting wikilinks/frontmatter/deploy glue.

trip2g lead-with. The whole conversion pipeline disappears: wikilinks are native (Goldmark), frontmatter is the routing/config, sync replaces the export-convert-commit-deploy loop; git still available (vault served over git Smart HTTP at /_system/git), so scripted workflows survive.

Where Hugo legitimately wins. Enormous theme ecosystem, total control of output HTML, free static hosting, battle-tested at scale, no server. If the reader enjoys the build pipeline, Hugo is the more flexible machine.

4. trip2g-vs-notion-sites.md — "notion sites alternative self-hosted"

Models.

• selfh.st — Self-Hosting Guide to Alternatives: Notion — the authority hub for "self-hosted alternative to X" queries (fetch blocked us, but it consistently tops these SERPs; study in browser).
• Docmost — 5 Open-Source Alternatives to Notion — vendor listicle: consistent per-tool shape (1–2-sentence intro, 5–7 feature bullets, GitHub link), vendor first, ~1,200 words, single CTA. Efficient but saltier than Unmarkdown; copy the section shape, not the self-first placement.
• openalternative.co — Open Source Notion Alternatives — programmatic-SEO directory shape; shows what we're up against for the head term (go long-tail: "notion sites alternative").

Why they rank. selfh.st = topical authority + audience trust; docmost = clean scannable structure + high-intent keyword; openalternative = programmatic scale.

Searcher intent. Notion user who liked one-click "Sites" publishing but wants out of per-seat pricing/lock-in, or needs self-hosting for compliance.

trip2g lead-with. Your content is markdown files you can git clone out at any moment (vs Notion export lossiness); self-hosted, no per-seat price; Obsidian as the editor; forms cover the "collect input" use Notion Sites is often chosen for; multi-domain.

Where Notion legitimately wins. WYSIWYG editing for non-technical teammates, databases/boards as first-class objects, zero infrastructure, real-time collaboration. If the team won't touch markdown, Notion is the right call — say it.

5. trip2g-vs-gitbook.md — "gitbook open source alternative"

Models.

• Docmost — Top 5 GitBook Alternatives — dissected above: ~2,000 words, vendor first with screenshot + feature list + CTA, competitors get pros/cons only. Works via primacy, but its bias is visible; blend its structure with Unmarkdown's honesty.
• apidog — Top 10 Best GitBook Alternatives and AlternativeTo — Open Source GitBook Alternatives — the listicle/directory incumbents; confirm the term is contested, so our page should target the narrower "gitbook alternative self-hosted"/"open source gitbook alternative for Obsidian".

Searcher intent. Team docs shopper: outgrew GitBook free tier or wants self-hosting.

trip2g lead-with. MIT self-hosted, one Go binary on SQLite (no Postgres/Redis stack to run — contrast even with other self-hosted wikis); authors write in Obsidian, not a web editor; custom Jet templates for branded docs; MCP means the docs are directly queryable by the team's AI agents — a row no GitBook alternative fills; git access to the whole base.

Where GitBook legitimately wins. Polished hosted UX, change-request/review workflow, API-docs tooling (OpenAPI), enterprise SSO tiers, zero ops. For a large team wanting managed docs with review gates, GitBook earns its price.

6. publish-obsidian-vault-as-website.md — head how-to: "how to publish an obsidian vault as a website"

Models.

• Bryan Hogan — Making a Website with Obsidian — the best head-term shape: answer-first opening, immediate method/complexity/cost table (Publish / Quartz / Astro Starlight / Custom), then 2–4 paragraphs per method, live example sites as proof. ~700–800 words.
• Obsidian Help — Set up Obsidian Publish — the official incumbent; our page must acknowledge it as method #1 to be credible.
• dev.to — Host your Obsidian notebook on GitHub Pages for free — dev.to ranks fast for dev how-tos; also the cross-post channel per growth-plan Step 4.

Why they rank. Bryan Hogan covers all methods for a query where every other page covers one; the options table matches "which way should I do this" intent exactly.

trip2g angle. Write it as an honest survey (Publish → Quartz → Hugo/Astro → trip2g), each with complexity/cost/best-for, then a full walkthrough of the trip2g path (install → connect vault → sync → live site) with a verification step. Favorable framing = trip2g is the only "live server" row (two-way sync, paywalls, MCP, forms); honest framing = it's also the only row that requires running a server (or the free cloud instance).

7. digital-garden-self-hosted.md — "digital garden self-hosted", "obsidian digital garden"

Models.

• Ness Labs — How to set up your own digital garden — the canonical intro that owns the concept query: defines digital garden, then no-code vs technical paths.
• Maggie Appleton — Digital Gardening for Non-Technical Folks — the community's most-linked author on the topic; audience-segmented structure (by skill level) worth copying.
• Flowershow — Self publish your digital garden — a docs-site tutorial ranking for the "self-hosted" modifier, i.e. exactly our slot.

Searcher intent. Concept-first, then tool-shopping; higher-funnel than the pages above.

trip2g lead-with. Gardens are tended continuously — two-way sync makes editing be publishing; monetization lets a garden have paid beds (unique among garden tools); RSS + Telegram give the garden distribution; agents can tend it over MCP.

Where competitors legitimately win. Static generators + GitHub Pages = $0 and zero ops (the default recommendation in every model article); Obsidian Publish = fastest no-code path. Position trip2g for the gardener who wants the garden to do things.

8. mcp-memory-server.md hub — "MCP server memory", "persistent memory MCP server", "self-hosted agent memory"

Models.

• ChatForest — Best Memory & Knowledge MCP Servers and Atlan — Best AI Agent Memory Frameworks — the roundup shape that currently owns these SERPs: ranked list, per-tool capsule, "no single best — depends on use case" verdict.
• PulseMCP directory (memory query) — directories rank for these terms; being listed (growth plan Step 3) matters as much as the page.
• Mem0 — OpenMemory — vendor landing that ranks by owning a sub-category name; supports our "own a dimension" move.

Why they rank. Thin competition (per seo_plan.md §1), exact-match titles, scannable capsule-per-server format.

trip2g lead-with. The memory is human-readable markdown notes — auditable, editable in Obsidian, versioned (note_versions + git mirror); token-efficient retrieval (search → expand → read_section; the 15–37× benchmark on token-economy-bench.md); SQLite single binary, no vector-DB stack; the same memory is a browsable website; federation fans a memory query across hubs.

Where competitors legitimately win. Purpose-built memory servers (ai-memory, agentmemory, OMEGA, Engram per the roundups) do automatic capture/scoring/promotion with zero authoring discipline; Mem0 cloud is zero-setup. trip2g memory is deliberate notes, not automatic conversation capture — say that; it's also the feature for users who want to read their agent's memory.

9. claude-code-persistent-memory.md — "Claude Code memory MCP", "claude code persistent memory"

Models.

• Mem0 — Add Persistent Memory to Claude Code (5-Minute Setup) — the benchmark vendor tutorial (dissected in TL;DR): friction intro, 3 install paths, verification, 8-Q FAQ, ~2,500 words.
• dev.to — How to give Claude Code persistent memory with a self-hosted mem0 MCP server — ~3,500 words; "self-hosted" appears 12+ times (deliberate modifier density); opens with a real debugging scene; discloses the LLM-call cost of graph mode and mitigations. Our closest intent match.
• MemPalace — How to Add Persistent Memory to Claude Code — small-vendor page ranking on exact-match title + year, proving the term is winnable.

Searcher intent. Claude Code user tired of re-explaining the project every session. One of the top community questions (per seo_plan.md).

trip2g lead-with. One MCP URL (/_system/mcp) — no local process per machine; memory shared across machines/teammates (subscription-scoped); memcli/quickstart; memory you can open in Obsidian and edit; token-economy numbers as the differentiator stat (the dev.to model shows cost disclosure is a feature).

Where competitors legitimately win. Anthropic's official memory MCP is free, local, zero-account; Mem0 auto-extracts facts without the agent being told to write notes; MemPalace stores conversations verbatim. If the user wants fully-local single-machine memory, the official server is simpler — recommend it for that case.

10. obsidian-mcp-server.md — "Obsidian MCP server"

Models.

• MCPBundles — Obsidian MCP Server: Connect Your AI to Your Vault — ~2,800 words; hook = "most Obsidian MCP servers give you six tools… this one does things the others can't"; 7×4 feature matrix vs three named competitors; 3-step setup; 6-question FAQ. The best vendor model in this niche.
• Awesome Claude — 3 Ways to Use Obsidian with Claude Code — ~800 words; three methods (MarkusPfundstein REST-API server / Anthropic Filesystem MCP / cyanheads plugin) ranked by GitHub stars with a stars/best-for/pros-cons table. The neutral survey shape.
• MarkusPfundstein/mcp-obsidian — the incumbent README that actually ranks; whatever we write must position against it accurately.

Searcher intent. Obsidian user wanting Claude/agents to read (and write) the vault.

trip2g lead-with. The existing servers all require the laptop on and Obsidian (or a local process) running; trip2g is the hosted variant — vault synced to a hub, agents query it from anywhere, access scoped by subscription/ACL, plus hybrid full-text + semantic search and version history. Also the honest "4th way" framing: extend Awesome Claude's 3-way survey with the server-hosted option.

Where competitors legitimately win. Fully local (nothing leaves the machine), no server to run, 2-minute setup, direct file access. For a single user on one machine, a local MCP server is genuinely simpler — recommend it for that case and link it.

11. trip2g-vs-mem0-vs-memgpt.md — agent-memory comparison shoppers

Models.

• Particula Tech — Agent Memory Frameworks Tested — the gold standard (see TL;DR): decision matrix, benchmark citations, per-tool "pick when", named weaknesses, "run your own eval" advice.
• TokenMix — Mem0 vs Letta vs MemGPT and dev.to — AI Agent Memory in 2026 — the crowded "vs" field; note several pieces already add a small newcomer to the title (e.g. Rohit Raj adds MemPalace) — the exact move for us: "Mem0 vs Letta vs Zep vs trip2g".
• Architecture one-liners to reuse from the SERP consensus: "Mem0 is a memory layer you bolt on; Letta is a runtime where the agent is its memory; Zep builds a temporal knowledge graph."

Searcher intent. Developer choosing memory infrastructure; comparison-shopping with benchmarks.

trip2g lead-with. A fourth architecture: "memory as a markdown knowledge base" — human-readable/editable, git-versioned, served to humans as a website and to agents over MCP; no extraction pipeline or vector DB to operate; token-economy benchmark as our quantifiable stat (the genre runs on numbers — LongMemEval cites are what make Particula's piece memorable).

Where competitors legitimately win. Mem0: automatic fact extraction, huge community (~47k stars), managed cloud. Letta/MemGPT: agent self-managed memory hierarchy, research pedigree. Zep: temporal reasoning / point-in-time correctness (Graphiti, 63.8% LongMemEval). trip2g doesn't auto-extract from conversations and has no LongMemEval score — state both; our "pick trip2g when" line: you want memory you can read, edit in Obsidian, and publish.

---