Content · App Starter
Knowledge Base Template
A documentation site with search, versioning, and feedback signals — generated into a codebase that answers to you.
Documentation dies in two ways: nobody can find it, or nobody trusts what they find. This starter generates a knowledge base that fights both — a searchable doc tree with a draft/publish gate, per-article version history, and helpful votes that tell you which pages are earning their keep. All of it ships as source code in your repository.
- 6 core entities
- Full-text search
- Code you own
- Vibe + Governed modes
01 · Data model
What's inside: the data model
The doc tree, the editorial workflow, and the feedback loop all read from six tables. Because they're plain relational entities in your schema, a docs migration is an import script — and a custom structure is a column, not a support ticket.
| table | what it holds |
|---|---|
spaces | Top-level doc collections — product docs, help center, internal handbook — each with its own navigation tree and access rules. |
articles | The pages themselves: title, slug, body, status, category, and position in the tree, with draft and published states. |
article_versions | An immutable snapshot per publish, so any page can be diffed against its history or rolled back in one step. |
categories | The tree's branches — nested sections that give a space its sidebar and keep two hundred articles navigable. |
feedback | Helpful/not-helpful votes with optional comments, tied to the article version the reader actually saw. |
search_queries | What readers searched and whether anything matched — the zero-result log is your content backlog, written by your users. |
02 · Screens
Signature modules
Public doc tree with search
01Readers get a sidebar tree, breadcrumbs, and full-text search across every published article. Search runs against your own database, so tuning what ranks — exact titles first, archived pages last — is your code, not a vendor's relevance black box.
Draft and publish workflow
02Articles start as drafts, visible to editors only, and go live on an explicit publish. Edits to a live page happen in draft too, so a half-finished rewrite never leaks into search results while the author is at lunch.
Version history
03Each publish snapshots the article. When a policy page said something different in March, you can prove exactly what — and when a rewrite makes a page worse, rollback is a click rather than an archaeology project.
Feedback and search analytics
04Helpful votes accumulate per version, and search queries log whether they matched anything. Together they answer the two questions doc teams guess at: which pages fail readers, and what do readers want that doesn't exist yet.
03 · Process
How it works
- 1You
Describe what you're documenting
Product docs, a customer help center, an internal handbook — tell Dual7 the audience, the structure, and who gets to publish. The starter shapes spaces, the tree, and permissions around that.
- 2Vibe mode
Refine the working site in Vibe mode
Adjust conversationally: add a reviewed-by field to articles, restrict a space to signed-in staff, change how search ranks titles. Every refinement is a readable change in your codebase.
- 3Governed mode
Publish via the governed pipeline
Deploy to your own infrastructure through approval gates, audit trail included. Your documentation — often your most-read content — lives on domains and databases you control.
04 · Fit
Who the Knowledge Base starter is for
Support leads deflecting repeat tickets
The same eight questions generate half the queue. A searchable help center with honest analytics shows whether the fix article actually deflects — and the zero-result search log tells you which ninth question to write up next.
Teams whose wiki became a landfill
Wikis rot because anyone can publish and no one can prune. A draft/publish gate plus per-page feedback flips the incentive: pages earn their place with helpful votes, and the ones that don't are visible candidates for deletion.
Companies keeping docs off third-party clouds
Internal handbooks hold org charts, processes, and sometimes secrets. Running the knowledge base on your own infrastructure — with access rules in code you audit — is a shorter conversation with security than any vendor questionnaire.
Product teams docs-as-a-feature
For developer tools and APIs, documentation is the product surface users touch most. Owning the docs app means the docs can do product-specific things — live examples, version pickers, generated references — that a generic hosted wiki will never ship for you.
05 · Yours
Customize it, own it
Make it yours
Structure is the usual fight with docs tools, and here it's just schema. Want an owner and a review-due date on every article? Two columns and a dashboard. Want per-space theming so the help center and the internal handbook stop looking identical? The templates are in your repo. Want approvals before publish? Add a reviewer step to the workflow — Vibe mode edits, all of them.
The generated app is deliberately boring underneath: articles are rows, search is a query, rendering is a normal web stack. That makes the exotic stuff tractable — embed live API calls in a page, pull reference sections from your codebase at build time, or mirror a space to PDF for the compliance binder. No plugin marketplace required, because there's no platform in the way.
Why a starter, not a subscription
Hosted docs platforms charge per editor and hold the content in their structure — export gives you files, but the tree, the history, and the feedback data stay behind. Generating the knowledge base as code turns that upside down: articles, versions, votes, and search logs are rows in your database, and the site serving them is an app in your repository with no meter on it.
The honest cost is stewardship — you deploy it, you back it up. Dual7's governed pipeline carries most of that weight with gated releases and an audit trail, and a knowledge base is a forgiving first tenant: read-heavy, low-write, and tolerant of a restart nobody notices.
06 · Industries
Industry versions
Knowledge Base template — frequently asked questions
What does the knowledge base starter include?
A full documentation app in code you own: spaces, articles with draft/publish states, immutable version history, nested categories, helpful-vote feedback, and logged search queries — plus a public reader site and an editorial admin over the same six tables.
Can it serve both a public help center and a private wiki?
Yes. Spaces carry their own access rules, so one deployment can serve a public help center and a staff-only handbook side by side. Private spaces require sign-in and stay out of public search and sitemaps.
How does search work — is it any good?
Full-text search runs against your own database, scoped to published articles the reader can access. Because ranking is code in your project, you can boost titles, bury deprecated sections, or add synonyms your users actually type — and the zero-result log shows where it still falls short.
Does it keep old versions of pages?
Every publish writes a snapshot to article_versions. You can diff any two versions, roll back a bad rewrite, and — because feedback links to the version a reader saw — tell whether that rewrite made a page more or less helpful.
Can I migrate existing docs into it?
The schema is plain relational tables, so a migration is an import script: walk your current export, create categories, insert articles, publish. Messy source formatting is the only hard part, and owning the code means you can fix it in the importer rather than by re-typing pages.
What about SEO for a public help center?
Published articles render server-side with clean URLs, titles, and descriptions from fields you control, and only published pages are indexable. Since the app is yours, adding structured data or custom canonicals is an edit, not a feature request.
Related
Dual7 App Starters
Build your Knowledge Base on your terms
A documentation site with search, versioning, and feedback signals — generated into a codebase that answers to you. Describe your version to start — the output is a project you own.