Skip to content
Back to search

obsidian-wiki

Framework for AI agents to build and maintain a digital brain through Obsidian wiki | Memory System for Agents

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026. Its scripts were only partly scanned.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/Ar9av/obsidian-wiki with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/Ar9av/obsidian-wiki'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/ar9av-obsidian-wiki/badge.svg)](/listings/ar9av-obsidian-wiki)

HTML

<a href="/listings/ar9av-obsidian-wiki"><img src="/api/listings/ar9av-obsidian-wiki/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 19 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. {
  2. "name": "obsidian-wiki",
  3. "owner": {
  4. "name": "Ar9av",
  5. "url": "https://github.com/Ar9av"
  6. },
  7. "metadata": {
  8. "description": "obsidian-wiki: an AI-maintained Obsidian knowledge base for Claude Code."
  9. },
  10. "plugins": [
  11. {
  12. "name": "obsidian-wiki",
  13. "source": "./",
  14. "description": "Grow an Obsidian second brain with your AI agent: ingest docs, chat history and URLs into interlinked markdown, then query it with citations.",
  15. "category": "productivity",
  16. "homepage": "https://github.com/Ar9av/obsidian-wiki"
  17. }
  18. ]
  19. }
README.md · README · 284 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <h1 align="center">obsidian-wiki</h1>
  2. <p align="center"><b>A unified digital brain you grow with your AI agents.</b></p>
  3. <p align="center">
  4. It remembers what you figure out, connects it to what you already know,<br>
  5. and answers when you ask.
  6. </p>
  7. <p align="center">
  8. <a href="https://pypi.org/project/obsidian-wiki/"><img src="https://img.shields.io/pypi/v/obsidian-wiki?color=blue" alt="PyPI" /></a>
  9. <a href="https://github.com/ar9av/obsidian-wiki/pulls"><img src="https://img.shields.io/badge/PRs-welcome-brightgreen.svg" alt="PRs Welcome" /></a>
  10. <a href="https://x.com/_ar9av"><img src="https://img.shields.io/badge/@__ar9av-black?logo=x&logoColor=white" alt="X" /></a>
  11. <a href="https://discord.gg/FH2PRX754c"><img src="https://img.shields.io/badge/Discord-join-5865F2?logo=discord&logoColor=white" alt="Discord" /></a>
  12. </p>
  13. <p align="center">
  14. <img width="768" alt="obsidian-wiki" src="assets/hero.png" />
  15. </p>
  16. <p align="center">
  17. <img width="768" alt="obsidian-wiki CLI: query, lint and context-pack on a demo vault" src="assets/demo.gif" />
  18. </p>
  19. <p align="center">
  20. English | <a href="https://github.com/Ar9av/obsidian-wiki/blob/main/README_TW.md">繁體中文</a>
  21. </p>
  22. ---
  23. You solve a hard problem on a Tuesday. Three months later, in a different repo, you solve it again from scratch — because the answer lived in a chat log you'll never find.
  24. This fixes that. Point it at a folder, tell your agent what to remember, and it compiles what you learn into interconnected markdown you own. The pattern comes from Andrej Karpathy's [LLM Wiki gist](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f): compile knowledge once and keep it current, instead of asking an LLM the same questions forever or re-running RAG every time.
  25. **Your second brain. Your AI agent is how you grow it.**
  26. Every skill here is a markdown file that any agent — Claude Code, Cursor, Codex, Windsurf, Gemini CLI, and [a dozen more](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/agents.md) — reads and runs. No runtime, no API keys, no vendor.
  27. ## 60 seconds
  28. ```bash
  29. pip install obsidian-wiki
  30. obsidian-wiki setup --vault ~/brain
  31. ```
  32. That also registers the two session hooks, so every new session starts with who you are and what's open, and ends by capturing anything reusable. Pass `--no-hooks` to skip them.
  33. Using `uv` or `pipx`? `uv tool install obsidian-wiki` and `pipx install obsidian-wiki` work the same way. (Not `uvx` — see [Installation](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/installation.md#install-via-pip-uv-or-pipx-recommended).)
  34. Then open any project in your agent and say **"set up my wiki"**.
  35. Prefer not to touch a terminal? Give your agent this and it'll do the whole thing:
  36. ```text
  37. https://github.com/Ar9av/obsidian-wiki — set up my wiki
  38. ```
  39. Using Claude Code? Install it as a plugin instead — no Python needed:
  40. ```text
  41. /plugin marketplace add Ar9av/obsidian-wiki
  42. /plugin install obsidian-wiki@obsidian-wiki
  43. ```
  44. Plugin skills are namespaced (`/obsidian-wiki:wiki-ingest`), and the plugin ships the skills only — not the session hooks or the `obsidian-wiki` CLI.
  45. Other paths — `git clone`, Claude Code plugin, Skills CLI, multiple vaults → **[Installation](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/installation.md)**
  46. ## What you actually do
  47. <p align="center">
  48. <img width="760" alt="The loop: feed it, it compiles into linked pages, ask it, keep it honest; every session starts from what the last one left behind" src="https://raw.githubusercontent.com/Ar9av/obsidian-wiki/main/assets/loop.svg" />
  49. </p>
  50. **Start with three.** `/wiki-setup` once, `/wiki-capture` to save what you just worked out, `/wiki-query` to get it back. Everything below is there when you need it.
  51. **Feed it.** Anything text-shaped: docs, PDFs, chat exports, meeting transcripts, screenshots, URLs.
  52. ```text
  53. /wiki-ingest ~/research
  54. /wiki-update # distill the repo you're standing in (code-graph aware)
  55. /wiki-capture # save this conversation
  56. /wiki-history-ingest claude # mine everything you've ever asked Claude
  57. ```
  58. **Ask it.** Answers come back with `[[wikilink]]` citations, not vibes.
  59. ```text
  60. /wiki-query what do I know about rate limiting?
  61. /wiki-narrate MCP security # a cited briefing on a topic
  62. /wiki-digest week # what did I learn this week?
  63. ```
  64. **Find that session you can't name.**
  65. ```bash
  66. obsidian-wiki sessions-build
  67. obsidian-wiki sessions-query "the auth bug with the weird retry loop"
  68. ```
  69. **Keep it honest.** The vault gets messy on its own; these clean it.
  70. ```text
  71. /wiki-lint # broken links, orphans, contradictions
  72. /wiki-dedup # "RSC" and "React Server Components" are one page now
  73. /cross-linker # weave new pages into the graph
  74. /wiki-status # what's ingested, what's pending, where the hubs are
  75. ```
  76. All 39 skills → **[Skills Reference](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/skills.md)**
  77. ## See it
  78. Open the vault in Obsidian and hit the graph view (Cmd/Ctrl+P → "Open graph view"). Say **"color my graph"** and it tints nodes by tag, category, or visibility.
  79. Want to look before you build? [`examples/demo-vault/`](https://github.com/Ar9av/obsidian-wiki/tree/main/examples/demo-vault) is an 18-page sample vault on building reliable LLM agents, already colored by folder. Clone the repo, choose *Open folder as vault* in Obsidian, and pick that folder.
  80. <p align="center">
  81. <img width="900" alt="obsidian-wiki graph view" src="https://github.com/user-attachments/assets/f2980840-4b5b-438a-8264-5ad1de42f483" />
  82. </p>
  83. Or export the whole graph to `graph.json`, GraphML (Gephi/yEd), Neo4j Cypher, Postgres SQL, or a self-contained interactive `graph.html`.
  84. ## Browse it in your browser
  85. No Obsidian needed. The vault comes with a local web viewer:
  86. ```bash
  87. pip install 'obsidian-wiki[server]' # Python 3.10+
  88. obsidian-wiki ui # opens http://127.0.0.1:3114/ui
  89. ```
  90. It shows an overview, search, rendered pages with backlinks and properties, the link graph, the activity log, your profile and open threads, the hot cache, and vault health. It listens on loopback only. Use `--vault` to choose a vault and `--port` to change the port.
  91. <p align="center">
  92. <img width="900" alt="obsidian-wiki web viewer: overview" src="https://github.com/Ar9av/obsidian-wiki/blob/main/docs/images/ui-overview.png?raw=true" />
  93. </p>
  94. <p align="center">
  95. <img width="900" alt="obsidian-wiki web viewer: a rendered page with backlinks" src="https://github.com/Ar9av/obsidian-wiki/blob/main/docs/images/ui-page.png?raw=true" />
  96. </p>
  97. **Connect any agent to it.** Open **Connect an agent** in the sidebar:
  98. 1. Pick your agent (Claude Code, Codex, Cursor, Claude Desktop, Gemini CLI, VS Code, Windsurf, or any MCP client) and copy the config. Stdio needs no server running. HTTP points at a hosted vault.
  99. 2. Paste the memory instructions into the agent's `AGENTS.md` / `CLAUDE.md` so it actually recalls at session start and saves findings back.
  100. 3. Check it: ask the agent to call `memory_recap`.
  101. <p align="center">
  102. <img width="900" alt="obsidian-wiki web viewer: connect an agent" src="https://github.com/Ar9av/obsidian-wiki/blob/main/docs/images/ui-connect.png?raw=true" />
  103. </p>
  104. More → **[Web viewer](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/deployment.md#web-viewer)**
  105. ## Use it from Python
  106. ```python
  107. from obsidian_wiki import Memory
  108. memory = Memory("~/brain")
  109. memory.remember("stack", "Python, FastAPI", confidence=0.9)
  110. memory.add("Postgres was chosen over MySQL for partial indexes.")
  111. memory.search("postgres")
  112. memory.recap() # inject at session start
  113. ```
  114. No API key, no embedding model, no vector store, no network — the vault is markdown on disk, so everything your agent remembers is readable, greppable and diffable. `user_id` scopes per-person memory while knowledge stays shared.
  115. More → **[Python API](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/python-api.md)**
  116. ## It remembers between sessions
  117. A session starts with the vault's memory already in context — who you are, what threads are open, what changed recently — and ends by capturing anything reusable before the context closes.
  118. `index.md`, `log.md`, `hot.md`, and the owner profile and todo index under `_meta/` all go through one writer that takes a lock and writes atomically, so parallel agents can't drop each other's updates.
  119. More → **[Memory Surface](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/memory.md)**
  120. ## Why this and not a notes folder
  121. - **It compiles, it doesn't accumulate.** New knowledge merges into existing pages. Contradictions get flagged. Nothing gets duplicated.
  122. - **It only reads what changed.** A manifest tracks every source ingested, so the second run processes the delta — not your whole library again.
  123. - **You can tell knowledge from guessing.** Every claim is tagged `extracted`, `^[inferred]`, or `^[ambiguous]`, and lint flags pages drifting into speculation.
  124. - **Queries stay cheap as it grows.** Titles, tags, and summaries get read before page bodies. 20 pages or 2000, roughly the same cost.
  125. - **It's yours.** Plain markdown in a folder. Push it to a private repo, open it in Obsidian, grep it, delete it. No service, no lock-in, nothing leaves your machine.
  126. - **It carries context across sessions.** An owner profile, open threads, and a running snapshot get injected at session start, so you don't re-explain yourself.
  127. - **Works where you already work.** One `skills/` directory, symlinked into every agent you use.
  128. **What it isn't**
  129. - **Not a transcript recorder.** Nothing is saved unless you or a skill decides it's worth keeping. The end-of-session hook only suggests a capture.
  130. - **Not a sync service or a backup.** It's a folder. Put it in git or your own sync.
  131. - **Not an oracle.** Pages are as good as their sources. Inferred claims are marked, and lint flags pages drifting into guesswork.
  132. - **Not a file converter.** Your agent reads text, PDFs, and images itself. Audio and video are skipped, and Office files depend on your agent. [What each input actually gets](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/architecture.md#what-each-input-actually-gets)
  133. More → **[Architecture](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/architecture.md)**
  134. ## How it compares
  135. | | obsidian-wiki | Desktop LLM-wiki apps | MCP memory servers | Single-skill LLM wikis |
  136. |---|---|---|---|---|
  137. | **Where knowledge lives** | Markdown in your Obsidian vault | A folder the app manages | A database or event log | A markdown folder |
  138. | **Works with** | Claude Code, Codex, Cursor, Gemini CLI, and a dozen more | The app itself | Any MCP client | Usually one agent |
  139. | **What you install** | Markdown skills; the Python CLI is optional | A desktop app plus an LLM API key | A server process | One skill |
  140. | **Mines past agent sessions** | Claude, Codex, Copilot, Hermes, OpenClaw, Pi | No | Captures new sessions only | No |
  141. | **Keeps itself clean** | Lint, dedup, cross-linking, tag taxonomy | Varies | Not applicable | Lint, if anything |
  142. ## Does it actually help?
  143. Structural questions — "how is X connected to Y", "which pages hold my vault together",
  144. "what breaks if I delete this" — are the ones a plain agent is worst at. It has to grep
  145. every file and reconstruct the link graph by hand, every single time.
  146. Same model, same vault, same questions. The only difference is whether `obsidian-wiki`
  147. was installed:
  148. | | Plain agent | With obsidian-wiki |
  149. |---|---|---|
  150. | **Time to answer** | 81s | **19s** — 4.4× faster |
  151. | **Correct answers** | 44% | **83%** |
  152. | **Tool calls used** | 9.9 | **4.6** |
  153. | **API cost** | $0.202 | $0.208 — unchanged |
  154. | Question | Plain agent | With obsidian-wiki |
  155. |---|---|---|
  156. | "How is X connected to Y?" | 122s | 18s |
  157. | "What topic clusters do I have?" | 117s | 21s |
  158. | "Which pages hold my vault together?" | 61s | 12s |
  159. | "What breaks if I delete X?" | 26s | 24s |
  160. The accuracy gap is not a rounding error. Asked to trace a connection, the plain agent
  161. routed through `index.md` — which links to *every* page, so it "found" a short path that
  162. means nothing. It made the same mistake in both runs, and named `index` as one of the
  163. most important pages in the vault. The graph the skills query excludes bookkeeping files,
  164. so that answer isn't reachable.
  165. <details>
  166. <summary>Method, and what this doesn't prove</summary>
  167. Claude Sonnet, headless, on a real 38-page vault. Questions were asked in plain English
  168. with no definition of the graph supplied — the plain agent had `Read`/`Grep`/`Glob`/`Bash`
  169. and had to work it out, which it did competently (it wrote its own centrality
  170. implementation rather than guessing). 4 questions × 2 conditions × 2 repetitions, run
  171. serially so nothing competed for CPU.
  172. Ground truth came from **networkx**, not from this project's own code: betweenness matches
  173. to 3.5e-18 across every node, and all 630 shortest-path pairs agree.
  174. It's a small study — n=2 per cell on one 38-page vault — so treat the exact percentages as
  175. indicative. The wall-clock gaps (3–6×) are much larger than the run-to-run spread; the
  176. accuracy figures rest on fewer samples. One run in the "with" column failed outright: the
  177. model ignored the CLI, grepped by hand, and got it wrong.
  178. Full data, per-run logs and the scaling measurements are in
  179. [PR #175](https://github.com/Ar9av/obsidian-wiki/pull/175).
  180. </details>
  181. ## Documentation
  182. | | |
  183. |---|---|
  184. | **[Installation](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/installation.md)** | pip, clone, agent-driven setup, multiple vaults |
  185. | **[Skills Reference](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/skills.md)** | All 39 skills and their slash commands |
  186. | **[Agent Compatibility](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/agents.md)** | The full matrix + per-agent manual setup |
  187. | **[CLI Reference](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/cli.md)** | Every `obsidian-wiki` subcommand |
  188. | **[Configuration](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/configuration.md)** | Config vars, QMD semantic search, `_raw/` staging, GitHub sync |
  189. | **[Architecture](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/architecture.md)** | The four ingest stages, vault structure, what we added to Karpathy's pattern |
  190. | **[Python API](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/python-api.md)** | `from obsidian_wiki import Memory` — use a vault as agent memory |
  191. | **[Memory Surface](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/memory.md)** | The index, log, hot cache, owner profile, and todo index |
  192. | **[Session Brain](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/session-brain.md)** | Topic graph over your agent session history |
  193. | **[Browser Extension](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/browser-extension.md)** | Capture pages into the vault, and fill web forms from it |
  194. | **[Deployment](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/deployment.md)** | The local web viewer, connecting agents over MCP, and self-hosting a vault (Docker or `obsidian-wiki serve`) with TLS, a read-only key and git sync |
  195. | **[Contributing](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/contributing.md)** | Adding skills, keeping the READMEs in sync |
  196. ## Contributing
  197. This is early. The skills work, but there's room to make the brain smarter — better cross-referencing, sharper deduplication, bigger vaults, new ingest sources. If you have a workflow that could be a skill, [PRs are welcome](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/contributing.md).
  198. Speak another language? Translating this README is a good first PR, and it needs no code: see [Adding a README translation](https://github.com/Ar9av/obsidian-wiki/blob/main/docs/contributing.md#adding-a-readme-translation).
  199. ## License
  200. [MIT](https://github.com/Ar9av/obsidian-wiki/blob/main/LICENSE)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.