Pages

Pages is JuhJuh's built-in knowledge base. Document your architecture, capture decisions, share runbooks. Every page you write feeds context into AI executions, and every successful execution makes that context stronger.

Your documentation works in two directions. Your team references it during onboarding, debugging, and planning. JuhJuh references it when executing tickets, choosing implementation strategies, and generating code that fits your codebase.

Creating pages¶

Open Pages from the sidebar and click New Page. The rich text editor supports markdown, code blocks with syntax highlighting, tables, task lists, and Mermaid diagrams. A live preview shows exactly what your team will read.

Every page supports metadata that keeps your knowledge base organized:

• Categories: Architecture, Domain, Pattern, Reference, Learning, Glossary, Business Rule, Workflow, Integration, Config, API Endpoint, Data Model, or Service
• Tags: color-coded labels for quick filtering. Create, rename, merge, and manage tags from the Tags view
• Sources: Manual (you write it), Execution (auto-generated from ticket work), Agent (created by AI during an execution), Sync (imported from external tools), or Wiki
• Icons: emoji icons on each page for visual scanning in the sidebar

Start typing [[ in the editor to insert a wiki link to another page. JuhJuh autocompletes the title and creates an internal cross-reference. These links stay valid even if the target page is renamed.

Templates¶

Page templates give you a head start on common formats. Open the Templates gallery to browse architecture decision records, runbooks, API documentation, onboarding guides, and more. Each comes pre-structured with the sections your team expects.

You can create your own templates, duplicate existing ones, and set a default template per folder. New pages created inside that folder start with the template's structure.

Auto-save and drafts¶

The editor auto-saves your work as you type. If you close the browser or navigate away, your draft is preserved. Return to the page and pick up where you left off. Click Discard Draft to revert to the last saved version.

Folders and organization¶

Group related pages into folders with up to three levels of nesting. A typical setup might look like Engineering / Backend / API Conventions. Deep enough to stay organized, flat enough to stay navigable.

Each folder carries its own settings:

• Default templates that apply to new pages created inside the folder
• Verification intervals that control how often pages surface for review (90 days by default)
• Visibility: Public (all organization members) or Private (restricted to folder members)
• Folder roles: Owner, Editor, and Viewer. These cascade to all pages within

Pin important pages to the top of any folder so they stay visible regardless of sort order. Star pages to bookmark them for quick access from your personal starred list.

When JuhJuh creates pages from ticket executions, it auto-routes them to the right folder based on category, ticket labels, and page title. A "Ticket Executions" folder is created automatically for execution-generated pages.

Bulk operations¶

Select multiple pages to move them between folders, change their category, or archive them in one action. Drag and drop to reorder pages and folders in the sidebar.

Version history¶

Every edit creates a new revision. JuhJuh tracks the full history of each page, so you can see how a document evolved over time.

View a side-by-side diff between any two revisions to see exactly what changed. If an edit introduces an error or removes critical context, restore a previous version with one click. Each revision records who made the change and when, giving your team a clear audit trail.

Permissions and sharing¶

Control who sees and edits each page with granular access levels.

Folder-level roles set the baseline:

Page-level permissions let you override the folder default for specific documents:

Share links let you distribute pages outside your organization. Each link has its own settings:

• Permission: view-only or view-and-comment
• Password protection: optional, set per link
• Expiration date: optional, auto-expires the link
• View count: tracks how many times the link has been accessed

Create multiple share links per page with different settings. Revoke any link at any time.

Comments and collaboration¶

Discuss pages without leaving the context. Threaded comments attach to specific sections of a page. Feedback stays anchored to the content it references.

• Resolution tracking: mark comments as resolved when the feedback has been addressed, and unresolve them if the discussion needs to reopen
• @mentions: tag team members by name to send them an in-app notification
• @juhjuh: ask AI a question directly in a comment thread, and JuhJuh responds with context-aware answers drawn from your knowledge base (see Ask JuhJuh)

Comments stay with the page through revisions, so the discussion history is always available.

Search¶

Find any page across your knowledge base with two search modes.

Full-text search matches keywords and phrases across page titles, content, tags, and categories. Filter results by folder, category, tag, author, date range, or verification status to narrow down large knowledge bases.

Semantic search finds pages by meaning. Search for "how we handle authentication" and JuhJuh surfaces pages about login flows, session management, and token refresh, even if those exact words never appear. Results include a relevance confidence score so you can gauge match quality at a glance.

Behind the scenes, JuhJuh combines eight ranking signals to surface the most relevant results: full-text match, semantic similarity, typo tolerance, confidence score, verification status, usage success rate, contextual relevance, and cross-reference weight. You do not need to tune any of this. The ranking adapts as your knowledge base grows.

Auto-learning knowledge base¶

Pages goes beyond a wiki. JuhJuh tracks which pages informed ticket executions and whether those executions succeeded. See Automation for how the feedback loop fits into the broader orchestration pipeline.

Confidence scoring. Each page carries a score between 0 and 1. When a page contributes context to a successful execution, its score rises. Scores decay over time for unused pages, keeping the signal fresh and the ranking honest.

Usage tracking. Every page shows which tickets referenced it and the outcome of each execution. You can see at a glance which documents are actively shaping your AI's decisions.

Stale detection. Pages not verified within their interval (90 days by default, configurable per folder) surface automatically for review. Stale pages do not disappear. They carry a visual flag that tells your team the content needs a human check.

Verification workflow. When you review a page and confirm its accuracy, click Verify to reset the clock. Verified pages regain their full confidence weight in future executions.

Duplicate detection. When creating a new page, JuhJuh checks for existing pages with matching content. If a near-duplicate exists, you can merge the two pages instead of creating redundant entries.

Analytics. The analytics dashboard shows your knowledge base health at a glance: total pages, stale count, pages not viewed in 60 days, verification rate, popular pages by view count, and unanswered questions that indicate knowledge gaps.

The more your team documents, the better JuhJuh executes. Your knowledge base is the context layer behind every AI decision.

Import and export¶

Import¶

Migrate your existing documentation into JuhJuh from Confluence or Notion. The importer preserves folder structure, page content, and metadata. A progress tracker shows the status of each page as it comes in.

Export¶

Export individual pages as Markdown, HTML, or PDF. Or export your entire knowledge base at once for backup or migration.

Integrations¶

Pages connects to Slack and Jira for cross-tool workflows:

• Slack: receive notifications in a channel when pages are created, updated, or verified
• Jira: link pages to Jira issues so your documentation stays connected to the work it describes

Configure these from the Pages settings within your project.

Documenting your work¶

Use Pages the way you'd use a team wiki, but with AI that actually reads what you write.

• Architecture decisions that inform how AI approaches refactors and structural changes
• API documentation that guides integration tickets so generated code follows your contracts
• Runbooks that shape incident-response executions with the steps your team already trusts
• Onboarding guides that help new team members and AI agents understand your codebase, conventions, and domain language

Start with the pages your team references most often. As your knowledge base grows, JuhJuh's executions get sharper. Each page gives JuhJuh more context for turning tickets into pull requests.

Related¶

• Automation: how Pages feed into continuous knowledge learning
• Tickets: where your knowledge base shapes AI execution
• Instructions: behavioral rules that complement documented knowledge
• Ask JuhJuh: get AI answers grounded in your Pages