
kepano/obsidian-skills
5 skills171k installs174k starsGitHub
Install
npx skills add https://github.com/kepano/obsidian-skillsSkills in this repo
1Obsidian Markdownobsidian-markdown is an agent skill for solo builders who keep specs, research, and runbooks in Obsidian vaults. It teaches valid Obsidian Flavored Markdown: YAML-style properties in frontmatter, double-bracket wikilinks for internal graph connections, embed syntax for notes images and PDFs, and typed callouts for highlights. Agents follow a six-step workflow from frontmatter through verification in reading view, with deeper rules deferred to bundled reference docs on properties, embeds, and callouts. Use it whenever you ask an agent to create or refactor .md files meant to open in Obsidian—not generic GitHub READMEs alone. It spans multiple journey phases because the same vault holds idea notes, build docs, and launch checklists.41kinstalls2Obsidian CliObsidian CLI is an agent skill for invoking the official Obsidian command-line interface against an open vault. Solo builders use it when they want agents to capture research, update specs, search notes, or automate vault maintenance without clicking through the GUI. Coverage spans note lifecycle (create with multiline `\n` content), file resolution via wikilink names or full paths, and vault scoping when multiple vaults are open. Developers extending Obsidian get operational commands for reloading plugins, executing JavaScript in-app, error capture, screenshots, and DOM inspection—useful when pairing agent coding with theme or plugin iteration. The skill points to https://help.obsidian.md/cli for full documentation while emphasizing `obsidian help` for an up-to-date command list. It is integration-focused: your agent shells out to Obsidian, so Obsidian must be running and the user must accept local CLI access on their machine.36kinstalls3Obsidian Basesobsidian-bases is an agent skill that exposes a structured reference for Obsidian Bases functions and types—parse dates, build links, coerce values, and compose conditional expressions inside Bases views. Solo and indie builders who run their specs, tasks, and research in Obsidian install it when they outgrow static folders and want spreadsheet-like queries over frontmatter and inline fields without leaving the vault. Use it while designing filters, rollups, and display formulas so your agent emits valid Bases syntax instead of guessing. It does not replace Obsidian core docs; it condenses callable signatures and field semantics for in-editor assistance. Best after you know what a Base is and which properties you want to surface; combine with broader Obsidian skills when you also need JSON Canvas, markdown, or CLI vault edits.35.3kinstalls4Json Canvasjson-canvas is an agent skill for solo builders who treat Obsidian as their operating system and need valid .canvas files instead of brittle one-off JSON. It encodes the JSON Canvas 1.0 model—nodes array, edges array, required geometry fields—and walks the agent through create, extend, and connect workflows with explicit validation after each change. Use it when you mention Canvas files, want a mind map in the vault, or need a flowchart that Obsidian can open natively. The skill emphasizes unique identifiers, non-colliding placement, and edge references that actually exist, which prevents silent breakage in sync’d vaults. It fits builders shipping content products, research notes, or agent-assisted PKM who want diagrams that stay editable and portable rather than screenshot-only artifacts.30.1kinstalls5DefuddleDefuddle is a small agent skill that wraps the Defuddle CLI to extract readable markdown from everyday web pages, stripping navigation, ads, and boilerplate so solo builders save context tokens when agents read articles, vendor docs, or blog posts. Install globally with npm when needed, then call defuddle parse with --md (or save with -o) instead of WebFetch whenever the user supplies a standard HTML URL. The skill explicitly tells agents not to use Defuddle for URLs that already end in .md—those should use WebFetch directly. Metadata flags (-p title, description, domain) help quick citations without pulling full page noise. It fits Obsidian-adjacent and general documentation workflows but is not a browser automation or paywalled scraper replacement. Intermediate complexity: you need Node/npm, network access, and judgment on which URLs are appropriate to fetch.29kinstalls