By waelmas
Maintain persistent project knowledge bases in Obsidian vaults for AI agents: initialize with codebase analysis and doc audits for staleness, scaffold structures, write/read session handoffs for continuity, refresh notes from git changes with evidence, and visualize note graphs.
Audit your project's markdown docs against the actual code - flag stale claims with file:line evidence
Open the project's Obsidian vault and show the knowledge graph view
Alias for /kb-offboard - write a session handoff note to the project's Obsidian knowledge base
Initialize the project knowledge base - scaffolds a vault (if missing) and populates it with rich docs + stale-doc audit via parallel subagents
Write a session handoff note to the project's Obsidian knowledge base
Audit the existing markdown documentation in the current project against the actual code. Flags stale claims with file:line evidence, categorizes each doc (current / partially stale / stale / aspirational), produces a single "Documentation Audit" note in the project's Obsidian knowledge base. Re-runnable at any time - use for quarterly doc refreshes or before a major rewrite. Use when the user asks to "audit our docs", "check for stale READMEs", "find outdated documentation", or types /kb-audit. Independent of kb-init's full populate flow.
Open the project's Obsidian vault in Obsidian and switch to the graph view, showing the knowledge base as a visual network of notes and their wiki-link relationships. Use when the user asks to "see the graph", "visualize the vault", "show me how notes connect", "open the graph view", or types /kb-graph. Also useful after kb-init to show off the newly rich vault.
Initialize the project knowledge base. If no vault exists, scaffolds one (via kb-scaffold) first. Then analyzes the codebase with parallel subagents and populates the vault with rich, structured documentation - mining existing markdown docs for insights but verifying every concrete claim against current code, so stale documentation doesn't pollute the vault. Produces curated notes plus a "Documentation Audit" report flagging outdated project docs. Use when the user asks to "initialize a knowledge base", "document this project", "populate the vault", "set up project memory", or types /kb-init. Always ask for confirmation before dispatching subagents - this uses meaningful compute and takes a few minutes.
Write a session handoff note to the project's Obsidian knowledge base so the next session can pick up seamlessly. Use when wrapping up a session, when the user says they're "done", "wrapping up", "that's it for today", or types /kb-offboard (even in environments without native slash command support - treat that as intent to trigger this skill). The handoff is adaptive - proportional to what actually happened in the session.
Brief the agent on what happened last session by reading the latest handoff note and project overview from an Obsidian knowledge base. Use when starting work on a project that has a knowledge vault, when the user asks to "catch up", "see what was done last time", "continue previous work", or types /kb-onboard (even in environments without native slash command support - treat that as intent to trigger this skill). Read-only - never modifies the vault.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Plugins for AI coding agents that fight context rot, doc rot, and forgotten decisions. Native on Claude Code, GitHub Copilot CLI, Cursor, and Codex CLI.
Built by Wael Masri.
Today, one plugin. More coming — they'll land as siblings in this repo.
obsidian-kb — persistent, per-project memory for your AI coding agentA real Obsidian vault per project, sitting inside the project folder by default (committable), holding structured knowledge your agent can read between sessions.
| Capability | Command | What happens |
|---|---|---|
| Build a project knowledge base from your codebase | /kb-init | Parallel subagents write Architecture / Tech Stack / Patterns / etc. notes into the vault. 3–6 minutes. |
| Keep the KB aligned with recent changes | /kb-update | Reviews git activity + session context, surgically refreshes stale KB notes with file:line evidence, flags new concepts as candidates. |
| Audit your project docs for drift | /kb-init or /kb-audit | Flags every stale claim in the project's own markdown (README.md, ARCHITECTURE.md, etc.) with file:line evidence. |
| Session handoff — beat context rot | /kb-offboard (alias /kb-handoff) | Writes an adaptive handoff into the vault (TL;DR, decisions, files changed, gotchas, next steps), linked from the index and wiki-linked to related KB notes. |
| Session onboarding — pick up in a new session | /kb-onboard | New agent reads the latest handoff and the linked KB notes before starting work. |
| Visual knowledge map | /kb-graph | Opens the vault in Obsidian's graph view. |
| Empty scaffold (hand-populated KB) | /kb-scaffold | Vault structure only, no codebase analysis. |
The vault is plain markdown on disk — committable alongside your code, editable by hand, grep-able, zero lock-in. Not a vector DB, not a CLAUDE.md replacement.
obsidian-kb needs two things:
brew install --cask obsidian / winget install Obsidian.Obsidian / download from obsidian.md.obsidian lands on PATH.Both are required. The plugin preflight tells you which is missing and walks you through enabling the CLI, but knowing upfront saves a round-trip.
Installation has two steps that live in two different places. Keep them straight or you'll paste the wrong command into the wrong place — which is the most common source of confusion.
1. Install the plugin — runs in your regular terminal (bash / zsh / PowerShell). 2. Use the plugin — runs inside your AI agent's chat or session, not in your terminal. Commands like
/kb-initare typed to the agent. They start with/because they're slash commands the agent recognizes; your shell doesn't.
Easiest — ask your agent to do it for you (from its own chat):
Install codeplow for me by following https://raw.githubusercontent.com/waelmas/codeplow/main/INSTALL.md
Or paste the one-liner into your terminal. It detects which AI tools you have installed and registers codeplow for each:
# macOS / Linux / WSL — paste into your terminal
curl -fsSL https://raw.githubusercontent.com/waelmas/codeplow/main/scripts/install.sh | bash
# Windows PowerShell — paste into a PowerShell window
iwr -useb https://raw.githubusercontent.com/waelmas/codeplow/main/scripts/install.ps1 | iex
All four snippets run in your regular terminal. They register codeplow with each tool's plugin system — they don't start the agent.
Claude Code
claude plugin marketplace add waelmas/codeplow
claude plugin install obsidian-kb@codeplow
GitHub Copilot CLI — reuses Claude Code's plugin format, so the same manifests serve both. Just a different binary:
copilot plugin marketplace add waelmas/codeplow
copilot plugin install obsidian-kb@codeplow
Cursor
git clone https://github.com/waelmas/codeplow ~/codeplow
mkdir -p ~/.cursor/plugins/local
ln -s ~/codeplow/obsidian-kb ~/.cursor/plugins/local/obsidian-kb
# Then in Cursor: Command Palette → Developer: Reload Window
Codex CLI
git clone https://github.com/waelmas/codeplow ~/codeplow
mkdir -p ~/.agents/plugins
cp ~/codeplow/.agents/plugins/marketplace.json ~/.agents/plugins/marketplace.json
# Then start Codex and open /plugins from inside its session
Once any of those finish, reload the agent so it picks up the plugin:
npx claudepluginhub waelmas/codeplow --plugin obsidian-kbBuild and maintain an LLM-curated personal knowledge base in your project — Andrej Karpathy's LLM Wiki pattern, designed to scale to thousands of pages without becoming a context bottleneck. Now with an optional compiled graph layer for typed, provenance-backed relationships.
Claude + Obsidian knowledge companion. Sets up a persistent, compounding wiki vault (Karpathy's LLM Wiki pattern). v1.7 "Compound Vault" + v1.8 methodology modes close 5 of 5 priority gaps from the May 2026 compass artifact. Ships: substrate alignment with kepano/obsidian-skills, default Obsidian CLI transport, hybrid retrieval (contextual prefix + BM25 + cosine rerank per Anthropic's Sept 2024 research), per-file advisory locking for multi-writer safety, pre-commit verifier agent, AND methodology modes (LYT / PARA / Zettelkasten / Generic) for first-class organizational support no other Claude+Obsidian competitor offers. v1.7.x audit closure: every BLOCKER + HIGH + MEDIUM + LOW finding from the v1.7.0 audit is CLOSED or DEFERRED-with-rationale. Optional DragonScale Memory extension (log folds, deterministic addresses, semantic tiling lint, boundary-first autoresearch).
Complete AI coding workflow system. Self-correcting memory + persistent FTS5-indexed research wikis + auto-research loop + multi-LLM council on a single SQLite store. 33 skills, 8 agents, 22 commands, 37 hook scripts across 24 events. Cross-agent via SkillKit.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
AI-powered wiki generator for code repositories. Generates comprehensive, Mermaid-rich documentation with dark-mode VitePress sites, onboarding guides, deep research, and source citations. Inspired by OpenDeepWiki and deepwiki-open.
Make your AI agent code with your project's architecture, rules, and decisions.