claude-obsidian-skills
Claude Skills for Obsidian Vault Management
Claude Obsidian Skills, reusable agent skills for Obsidian and Markdown
Claude Obsidian Skills is a collection of AI agent skills that automate personal knowledge management and Markdown workflows inside Claude Code, Gemini CLI, and similar assistants.
Skills included in the collection
The repository groups its skills into a few families. The PKM management group holds ai4pkm-helper, described as an onboarding guide and quick help tool that hands off to a daily use workflow, and gobi-onboarding, a voice based onboarding guide with a four step flow. The ai4pkm-helper offers onboarding guidance and quick help with a handoff to a daily use workflow, and gobi-onboarding is a voice based onboarding guide that follows a four step flow, so both target the early setup experience rather than long term maintenance. The Obsidian group is the largest and covers canvas, links, YAML frontmatter, markdown structure, and mermaid diagrams. The obsidian-canvas skill creates and manages Obsidian Canvas files with automatic layout generation for knowledge maps and visual summaries. The obsidian-links skill formats, validates, and fixes wiki links using proper file name, section, and folder conventions. The obsidian-yaml-frontmatter skill standardizes YAML frontmatter properties with consistent formatting and naming, while obsidian-markdown-structure validates heading hierarchy and document organization. The obsidian-mermaid skill builds Obsidian compatible Mermaid diagrams and warns about common errors such as markdown inside labels and layouts that are too wide. The Markdown group adds markdown-slides for Deckset and Marp presentations and interactive-writing-assistant for outline and prose co evolution. Video and image skills cover markdown-video, video-cleaning, and gemini-image-skill. The markdown-video skill turns Markdown slides into an MP4 video with text to speech narration, with or without Deckset, while video-cleaning removes silent pauses from videos using AI powered transcription and FFmpeg, and gemini-image-skill generates AI images through the Google Gemini API with several models and aspect ratios. Each skill is a self contained folder with its own SKILL.md, so you can copy only the ones you need rather than adopting the whole set.
Installing the skills for Claude Code and other assistants
The README gives a straightforward install path for Claude Code. You clone the repository, then copy the skill folders into your Claude Code skills directory with a command such as cp -r claude-obsidian-skills/*-* into the ~/.claude/skills folder, and Claude Code detects and uses them automatically. For other assistants the instructions are simpler: copy the SKILL.md file from the skill folder you want into that assistant's skill directory. Because each skill documents itself inside its SKILL.md, an assistant can load it without extra configuration. The README also describes an advanced option for vault maintainers who want to keep skills in sync. You can add the repository as a git submodule inside your vault and then create symlinks from the submodule's skill folders to your own skills directory, which gives you a single source of truth and version control. Updating is then a matter of running git submodule update with the remote flag. The note about recursive cloning matters here: when you clone a vault that contains submodules, you pass the recursive flag so the skills come down with it. This approach avoids duplicate copies of the skills across many vaults while still letting each vault reference the same files. The README also notes that when you clone a vault that contains submodules, you should pass the recursive flag so the skills are pulled down together with the rest of the repository rather than left empty.
Requirements and what each skill needs to run
Some skills need external tools and accounts before they work, and the README lists these in a requirements table. The markdown-video and video-cleaning skills both need FFmpeg, an OpenAI API key, Python 3.7 or newer, and Pillow. The gemini-image-skill needs a Gemini API key, Python 3.7 or newer, the google-genai package, and Pillow. The markdown-slides skill needs nothing beyond Markdown, and the obsidian skills need nothing because they only perform file operations. The interactive-writing-assistant also needs no extra dependencies. The common install commands gather the shared pieces: brew install ffmpeg on macOS, pip install Pillow openai google-genai for the Python libraries, and exported environment variables for OPENAI_API_KEY and GEMINI_API_KEY. The README also shows example prompts you can use to trigger a skill, such as asking the links skill to check and fix broken wiki links, asking the frontmatter skill to standardize properties, asking the mermaid skill to draw a flowchart, or asking the canvas skill to build a knowledge map. The skill structure section explains that every skill folder holds a SKILL.md plus optional reference and examples folders, and that the SKILL.md carries the metadata, the when to use guidance, the core rules, the step by step workflow, and a quality checklist. Contributors can add new skills by creating a folder with a SKILL.md that follows that same shape.
Editorial conclusion
Claude Obsidian Skills is an MIT licensed collection of agent skills that you install by copying skill folders into the ~/.claude/skills directory, where Claude Code detects and uses them automatically.
Community notes