Skip to content

phelps-sg/claude-code-obsidian-skills

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 

Repository files navigation

Claude Code + Obsidian Skills

A set of Claude Code skills that turn an Obsidian vault into a shared knowledge base for human and AI collaboration on software projects.

These skills are companion material for the blog post A Shared Memory for Claude Code: Using Obsidian to Organise Human and AI Knowledge.

What's included

Skill Invocation Purpose
Vault Manager /pkm Core read/write skill. Defines vault conventions and instructs the agent to write notes when discovering architecture, and to orient at the start of each session.
Vault Auditor /vault-insights Periodic audit that checks the vault for contradictions, broken links, and latent cross-cutting insights implied by the graph structure.
External Sync /external-sync Bridges an external knowledge source (e.g. Notion, Confluence) into the vault, with deduplication and human confirmation.
WIP Dashboard /wip Generates an Obsidian Kanban board from the vault, GitHub, and external project tools.

Installation

  1. Copy the skill directories into your Claude Code project's .claude/skills/ directory (or your global ~/.claude/skills/).
  2. Edit the vault path in each SKILL.md to point to your Obsidian vault.
  3. Adapt the tag taxonomy to your domain (the examples use #service/, #project/, #domain/ — replace with your own).
  4. For the External Sync skill, replace the Notion MCP tool references with whatever API your external source uses.
  5. For the WIP Dashboard skill, replace the GitHub repo references with your own.

Recommended CLAUDE.md configuration

The skills define how to read and write the vault, but you also need to tell the agent when to use it. Add the following to your CLAUDE.md (project-level or ~/.claude/CLAUDE.md for global):

# IMPORTANT: Vault before code

A shared Obsidian vault at `~/Documents/obsidian-vault/` serves as persistent
memory across sessions and projects. Use the `/pkm` skill to read and write
notes.

Before exploring code for any architectural, domain, or design question, ALWAYS
search the vault first. It contains accumulated context — architecture notes,
decision records, prior investigations, and domain knowledge — that prevents
redundant code trawling. Do NOT launch Explore agents or grep the codebase
until you have checked the vault for existing notes on the topic.

Why this matters: Without an explicit instruction, the agent will default to exploring the codebase directly. A vault with accumulated context is often faster and more comprehensive than ad-hoc code searches, but the agent needs to be told to check it first. Place this instruction near the top of your CLAUDE.md so it's read with highest priority. Mentioning the /pkm skill ensures the agent knows how to interact with the vault, not just that it exists.

Adapting to your stack

These skills were extracted from a real working setup and anonymised. You will need to customise:

  • Vault path: currently set to ~/Documents/obsidian-vault/
  • Tag taxonomy: the #service/, #lib/, #project/ hierarchy reflects a monorepo with multiple services — adapt to your repo structure
  • External tools: the sync and WIP skills reference Notion and GitHub — swap for your equivalents
  • File naming prefixes: pr-NNNNN-, decision-, lib- etc. — adjust to match your workflow

The vault manager skill is the foundation. Start there, get the conventions working, then add the others incrementally.

License

MIT

About

Claude Code skills that turn an Obsidian vault into a shared human/AI knowledge base

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors