The Birth of a "Meta-Tool"
Imagine you take notes in Obsidian—bidirectional links, graph view, Canvas, Bases (database views). Now you want an AI to help manage your vault. But when Claude Code opens it, it has no idea what a .canvas file looks like, nor does it understand the difference between [[wikilinks]] and plain Markdown links.
kepano/obsidian-skills fills exactly this gap. The author is none other than Obsidian's CEO, Steph Ango (known online as kepano). This isn't a third-party plugin from an enthusiast—it's Obsidian officially writing an "operations manual" for AI agents.
What Does It Provide?
Five skills, each targeting a data format or tool unique to Obsidian:
| Skill | Purpose |
|-------|---------|
| obsidian-markdown | Handles Obsidian-flavored Markdown—wikilinks, embeds, callouts, properties |
| obsidian-bases | Operates on .base files—Obsidian's database views, with filters, formulas, summaries |
| json-canvas | Operates on .canvas files—visual canvases in the JSON Canvas format |
| obsidian-cli | Invokes the Obsidian CLI—for plugin and theme development |
| defuddle | Extracts clean Markdown from web pages—removing noise to save tokens |
If you're unsure why these formats are special, think of it this way: plain Markdown is the "standard language" AI already speaks, while Obsidian's wikilinks, Canvas, and Bases are "dialects." Without learning the dialect, an AI treats [[Note A]] as plain text instead of a link, and sees a .canvas file as incomprehensible JSON.
The Agent Skills Specification: A Trend Worth Noting
The most significant thing about this project isn't any individual skill—it's that it follows the Agent Skills specification. The README states explicitly:
> These skills follow the Agent Skills specification so they can be used by any skills-compatible agent, including Claude Code, Codex, and Open Code.
In other words: these skills aren't tied to Claude Code. Codex, OpenCode—any spec-compliant agent can use the same skill files. Installation paths differ (Claude Code uses .claude/, Codex uses ~/.codex/skills/, OpenCode uses ~/.opencode/skills/), but the skill content itself is universal.
An apt analogy: skill files are to AI agents what drivers are to operating systems. Once every OS had to write its own drivers; then universal driver frameworks emerged. The Agent Skills spec is doing something similar—making "AI capability packages" reusable across agents.
Why Would the Obsidian CEO Write This Himself?
This signal matters more than the project itself.
Obsidian's business model has always been selling Sync and Publish services while keeping the client free and local. Now the CEO is personally writing Agent Skills, which shows Obsidian treats "AI operability" as infrastructure-level investment.
The reason is easy to guess: if AI agents can't correctly read and write Obsidian's file formats, users will migrate to more AI-friendly tools. Rather than waiting for third parties, better to have the official side define the standard. It's the same logic as database vendors writing their own ORM drivers—whoever controls the interface controls the ecosystem.
Defuddle: An Underrated Skill
Of the five skills, defuddle is the easiest to overlook but may deliver the most practical value. What it does seems simple: extract clean Markdown from web pages, stripping navigation bars, ads, scripts, and other noise.
But the keyword is "saving tokens."
When an agent needs to read a web page, feeding it raw HTML can burn 20K+ tokens on navigation and scripts for a complex page. Defuddle extracts the actual content, potentially needing only 2K tokens. For agents billed by token with limited context windows, this isn't a nice-to-have—it's real cost savings.
It's similar to the web-reader skill, but Defuddle is native to the Obsidian ecosystem and meshes seamlessly with Markdown notes.
Relationship to Anthropic Skills
Notably, anthropics/skills (Anthropic's official skill repository) was trending the same day. Both projects trending together is no coincidence—the Agent Skills ecosystem is forming rapidly.
The difference:
- anthropics/skills is a general-purpose skill collection maintained by Anthropic
- kepano/obsidian-skills is an app-specific skill set maintained officially by a third-party app (Obsidian)
- Repository: kepano/obsidian-skills
- +411 stars today
- Author: Steph Ango (Obsidian CEO)
- Specification: agentskills.io/specification
- License: not labeled (not specified in README)
The relationship resembles "standard libraries bundled with an OS" versus "SDKs provided by app vendors." You need the former to get an agent basically running, and the latter to let an agent operate specific tools.
Takeaways for Developers
If you're building a tool product, obsidian-skills offers a clear template:
1. If your data format isn't a standard one, it needs a skill—otherwise AI will treat it as unrecognized text 2. Follow the Agent Skills specification—don't lock into a single agent; give users choice 3. Official involvement is more convincing than third parties—users trust the vendor's understanding of its own formats
Data
*The emergence of the Agent Skills specification means the AI agent ecosystem is moving from "every agent doing its own thing" toward "portable capabilities." Obsidian's CEO personally writing skills is an early signal of app vendors proactively adapting to the agent ecosystem.*