Skill: obsidian-dataview-expert

What I do

I provide definitive expertise in writing Dataview queries (DQL) and JavaScript-based views (DataviewJS) within Obsidian. I enable agents to transform static knowledge bases into dynamic, self-organising databases by treating the vault as a queryable data source.

When to use me

• When creating or updating Obsidian Knowledge Base (KB) pages.
• When dynamic indexing of notes, skills, agents, or tasks is required.
• When building dashboards that must reflect the current state of the vault.
• When replacing static markdown tables with dynamic data views.
• CRITICAL RULE: Use me for ANY KB index page. NEVER use static markdown tables or manual lists in Obsidian KB pages. ALWAYS use DataviewJS queries that dynamically pull from vault metadata.

Core principles

1. Metadata-First Architecture: Treat frontmatter and tags as query fuel. No metadata means no visibility.
2. Defensive Programming: ALWAYS wrap DataviewJS in try/catch blocks with user-friendly error messages to prevent dashboard crashes.
3. Progressive Complexity: Use DQL for simple lists/tables; escalate to DataviewJS for complex logic, multi-step filtering, or custom CSS-styled rendering.
4. Path-Based Scoping: Narrow query scope using folder paths (e.g., startsWith("3. Resources/KB")) to ensure performance and accuracy.
5. British English: All labels, headers, and documentation within queries must use British English spelling.

DQL vs DataviewJS

DataviewJS fundamentals

Querying and Filtering

// Scoped query by path and tag
const base = "3. Resources/Knowledge Base/AI Development System";
const pages = dv.pages().where(p => p.file.path.startsWith(base));

// Tag matching (handling both single strings and arrays)
const skills = pages.where(p => 
    p.file.tags.values.some(t => t.startsWith("#skill/"))
);

Rendering Components

dv.header(2, "Active Skills");
dv.table(["Skill", "Category"], 
    skills.map(p => [p.file.link, p.category])
);

Common patterns

The Quick Stats Counter

Used for high-level dashboard summaries.

try {
    const pages = dv.pages("#type/note");
    const count = pages.length;
    dv.table(["Metric", "Count"], [
        ["Total Knowledge Assets", count]
    ]);
} catch (e) {
    dv.paragraph("⚠️ Error loading stats.");
}

The CSS Grid Skill Card

For visually engaging resource indexes (requires dashboard cssclass in frontmatter).

const groups = skills.groupBy(p => p.category);
for (const group of groups) {
    dv.header(3, group.key);
    dv.list(group.rows.file.link);
}

Error handling

MANDATORY TEMPLATE: Never write naked DataviewJS. Always use this wrapper:

try {
    // 1. Gather Data
    const data = dv.pages("#tag").where(condition);
    // 2. Process Data
    if (data.length === 0) {
        dv.paragraph("No matching resources found.");
        return;
    }
    // 3. Render Data
    dv.list(data.file.link);
} catch (e) {
    console.error("Dataview Error:", e);
    dv.paragraph("⚠️ Error rendering view. Check console for details.");
}

Anti-patterns to avoid

• ❌ Static Tables: Manual markdown tables in index pages. These go out of date instantly.
• ❌ Naked JS: DataviewJS without try/catch. This causes the entire page to break if a single note has malformed metadata.
• ❌ Vault-Wide Scoping: Using dv.pages() without where or FROM filters. This is slow and pulls irrelevant data.
• ❌ Hardcoded Values: Hardcoding dates or counts that should be derived from note metadata.
• ❌ American English: Using color instead of colour or initialize instead of initialise in labels.

KB Reference

~/vaults/baphled/3. Resources/Knowledge Base/AI Development System/Skills/Session-Knowledge/Obsidian Dataview Expert.md

Related skills

• obsidian-frontmatter: Source of truth for all Dataview queries.
• obsidian-structure: Defines the PARA paths used for scoped queries.
• british-english: Ensures consistency in all rendered dashboard text.
• obsidian-customjs-expert: For offloading complex logic to shared scripts.