gosidian

gosidian

Markdown notes your AI agents can read, write, and reason over — via MCP.

A self-contained markdown vault with a built-in MCP server. Humans edit through a web UI, agents talk to it over MCP, everything lives in plain .md files that Obsidian (and every other markdown tool) reads natively.

Listed on the official MCP Registry as io.github.daniele-chiappa/gosidian, Glama and mcpservers.org

Try it in your browser

Launch a free, throwaway gosidian in GitHub Codespaces — no install, running on your own Codespaces quota. It builds from source, seeds a small demo vault, and opens the web UI. Log in with demo / gosidian-demo.

Quick start

docker run -d --name gosidian \
  -p 8080:8080 \
  -v "$(pwd)/vault:/vault" \
  ghcr.io/daniele-chiappa/gosidian:latest
# open http://localhost:8080, create admin, copy the MCP token from /admin/tokens
claude mcp add gosidian http://localhost:8080/mcp \
  --transport http --header "Authorization: Bearer $TOKEN"

Three commands: Docker up → token created from the web UI → agent wired. Your .md vault is persisted under ./vault/; stop the container and the files are still there.

Other installation paths (source, custom compose, bare-metal): docs/getting-started.md.

What is gosidian

• A markdown vault. Notes are .md files on disk. Open the same folder in Obsidian, VS Code, vim, or any editor you already use. Zero lock-in: delete .gosidian/ and you have a pure Obsidian vault.
• An MCP server. 71 typed tools let agents bootstrap a session, ingest files, search, read, write, link, handoff, self-check, audit. Bearer tokens and OAuth grants, each narrowed on every request to what its account may read and write.
• A web UI. A Vue 3 single-page app served from the same binary (built with Vite, embedded via go:embed). Notes, graph, search and config forms open as windows in a tiling "plancia" workspace — full-text search, backlinks, graph view, editor with live preview, audit trail, admin pages for tokens and users.

All three views hit the same files on disk. The SQLite FTS5 index is a cache — drop it and it rebuilds.

Who it's for

• AI engineers wiring agents that need persistent structured memory: note-taking, plans, skills, ADRs, handoffs, audit.
• Obsidian users who want a programmable layer on top of a vault they already trust.
• Teams sharing one vault: per-project visibility and grants to accounts and teams, restricted new accounts with their own project, self-service tokens per agent that never outrun their owner.

Why gosidian instead of X

• vs RAG / vector search: gosidian retrieves the agents' memory by identity (path, tag, frontmatter, backlinks) — more predictable than similarity search for an agent's working memory. Semantic search over the notes is deliberately deferred: see ADR-007 rationale. A RAG over a project's documents (manuals, reports, scans) is another matter, and an experimental part of gosidian.
• vs Obsidian Sync: Sync mirrors a vault between human devices. gosidian adds a typed automation surface (MCP) to the same vault. Not competitive — complementary.
• vs Notion / Roam: hosted or proprietary formats; migration is a project. gosidian's vault is already .md files you can take anywhere.

FAQ covers the long form.

Compared to similar projects

The "markdown vault + agents" space is crowded. This is where gosidian sits and where it doesn't, as of September 2026:

If you are looking at…	What those projects do	Where gosidian differs
Obsidian MCP bridges — mcp-obsidian, obsidian-local-rest-api, wrappers around the Obsidian CLI	Expose a running Obsidian desktop app to agents over MCP.	Headless server: no Obsidian process needed, runs on a box or in a container, multi-user with roles, per-project visibility and grants (accounts and teams), tokens narrowed to each account, audit trail. The vault stays a plain Obsidian vault.
Markdown memory servers — basic-memory, mcp-vault	Same "files, not a database" idea; usually add hybrid semantic search, a cloud tier or WebDAV sync.	Ships a full web UI, real multi-user, an agent handoff bus and a server-served working method (versioned directives, lint, stale detection). No semantic search by design (ADR-007), no hosted tier.
Client-side wiki skills — obsidian-wiki and the "LLM Wiki" pattern	Slash-commands the agent runs locally to compile and maintain a wiki. No server, no auth, no UI.	The same pattern implemented server-side and agent-agnostic: one-call scaffold, directives served at bootstrap, handoffs, audit. Complementary: those skills work against a gosidian vault too.
Agent memory services — mem0, agentmemory, mcp-memory-service	Memory as an opaque store: embeddings, recall benchmarks, auto-capture hooks.	Memory is markdown that humans read in Obsidian or the web UI, retrieved by identity and graph, with no embeddings; a language model is called only by the pipelines you turn on. Auto-capture through Claude Code hooks (contrib/claude-code/: focus injected at session start, session digests at compaction and end) rather than inside the binary. No published recall numbers yet.
Note apps with community MCP servers — SilverBullet, Trilium, SiYuan	Mature editors, mobile apps, sometimes real-time collaboration; MCP added by third-party servers on top of their API.	MCP-first: the server is in the binary, behind the same login, roles and audit as the UI. No mobile app, no real-time collaboration, no WYSIWYG editor.

Honest gaps, in the order they come up: semantic / hybrid search over the notes (deferred, see the FAQ; documents have an experimental RAG), mobile sync beyond git, real-time collaboration, a hosted offering. The roadmap says which of these are planned.

Feature highlights

• Single binary, ≤50 MB, Alpine-based Docker image
• Web UI: a Vue 3 SPA (Vite, Pinia, Tailwind, CodeMirror, Cytoscape), embedded in the binary — editor + live preview, sidebar, search, graph view, attachments, audit log, admin pages
• Plancia tiling window manager (niri-style): notes, graph, search and config forms open as resizable, side-by-side windows in a horizontally-scrollable workspace, restorable from the URL
• MCP server over Streamable HTTP (legacy HTTP+SSE kept) with 71 typed tools
• Bearer tokens with scopes (read / write) and per-project restriction — including multi-project tokens for orchestrators; every token owned by an account is narrowed on each request to what that account may read and write, accounts mint their own (inherit follows their access, custom pins a subset), cascade-revoke on user disable
• Agent orchestration bus: handoff notes with an atomic claim/complete lifecycle, server-stamped identity, and a memory_wait_changes long-poll change feed — a minimal multi-agent task queue where everything stays plain markdown
• Multi-user web login with roles (Admin / User / Read-only) as the ceiling, per-project visibility (private / internal / public) for reading and grants (read / write / admin) to accounts and teams for everything else, delegated to project admins; new accounts start restricted with a personal project of their own; invite-only signup (24h TTL)
• Optional TOTP two-factor (global mode + per-user override) and LDAP / Active Directory login with guest auto-provisioning
• Opt-in OAuth 2.1 authorization server so claude.ai / Claude Desktop custom connectors, ChatGPT connectors and Claude Code's browser login get their own tokens through a consent screen — no pasted bearer
• Claude Code hooks (contrib/claude-code/): the project's hot.md focus injected at every session start, a digest of the session appended to the vault at compaction and at the end — no LLM, over POST /mcp/append, never blocking the session
• Optional git sync (debounced commits, push with token auth)
• SQLite FTS5 full-text search + ETag optimistic locking
• First-class .html notes, rendered in a sandboxed iframe (off by default, opt-in per project)
• Graph analytics over the wikilink graph: memory_hubs (most-linked notes) and memory_path (shortest path between two notes)
• Opt-in self-improve loop: agents record usage-friction insights per token, off by default
• Print / Save-as-PDF for any markdown note straight from the web UI
• Internationalization (IT + EN complete; ES / FR / DE scaffolding)
• Light & dark theme presets (Catppuccin, Tokyo Night, Solarized) + custom palette
• Opinionated Karpathy-Wiki-Stack project layout with one-call scaffolding
• Optional global projects for skills, agents & scaffold templates shared across projects (opt-in per project, local-overrides-global)

Experimental: pipelines, plugins, RAG and agents

These parts are new, off by default, and still changing from one release to the next: their settings, node types and notes may move, and docs/upgrading.md says when they do. The core above (the vault, the MCP server, the web UI) does not depend on them, and an install that leaves them off sees nothing of them.

• Pipelines (docs/pipelines.md): a note type: pipeline that the server runs when something happens — a file arrives, a commit, a chat message, a call from an agent or a service, the clock. Its nodes read and write notes, call a language model (any OpenAI-compatible endpoint, local or remote), run a script, or wait for a person's answer. One queue for all of them, with limits per plugin, provider and shared resource (a GPU, say), fair among the projects. Edited as text or in a visual editor. They run only once [pipelines] enabled = true and a project turns them on.
• Plugins (docs/plugins.md, gosidian-plugins): MCP servers beside gosidian, a container each, that add node types to the pipelines and tools to the agents. A plugin runs only after the owner accepts its manifest, and never sees the vault, only the files a pipeline hands it. Published so far:
  • documents: docling (PDF, office files, HTML to markdown), ocr (scans, through a vision model), images;
  • data: database (embeddings on PostgreSQL with pgvector, Qdrant or SQLite), storage (Nextcloud, S3 or Google Drive as a source of files);
  • people: chat (a bot as a widget for any web page), notify (Telegram, ntfy, email, with questions to answer);
  • agents and code: claude (Claude Code) and pi (pi.dev) as executors, exec (scripts in a sandbox), preview (a static build at an address no one guesses).
• The IN folder (docs/in.md): a project's files to work on, under their own names — uploaded from the web UI in resumable pieces, sent by an agent, or read from a remote store — which the pipelines turn into notes.
• RAG on a project's documents (docs/chat.md, docs/rag-benchmarks.md): the project template rag converts the files of the IN folder, cuts the notes into paragraphs, stores them with their embeddings, and answers a question from them, citing its sources; a bot (type: chat) asks it, on a page of its own or a widget on any site. Measured with local models on the eight public corpora of ViDoRe V3 (14,514 questions): search NDCG@10 52.7 on paragraphs, 57% of the answers judged good, a median 5.5 s per answer. The agents' own memory stays retrieval by identity: the RAG is for documents, not for the vault's notes.
• Teams of agents (docs/agents.md): a note type: team with its roles — a coordinator, writers, developers, reviewers — each an agent that a pipeline runs (Claude Code or pi, through their plugins), on tasks that are notes of the project. gosidian gives the tasks out, checks what each agent changed against its role's paths and checks, asks a person before a merge, and counts the cost against a budget of the day. A pilot runs the same team again and again on one objective: training plans for five people and a small web app.
• Compose presets (docs/deployment.md): deploy/presets/ starts gosidian with the RAG chain (docling, images, the database plugin on Qdrant, chat) from one .env.

Documentation

Area	Start here
Install + configure	Getting started, Configuration, Deployment, Upgrading
MCP integration	Overview, Tool catalogue, Authentication, Client setup, Agent patterns, Claude Code hooks
Web UI	Overview, Editor, Authentication & roles, Settings
Vault	Format, Conventions, Multi-project, Global projects, Obsidian compatibility
Experimental	Pipelines, Plugins, The IN folder, Bots, Teams of agents, RAG benchmarks
Internals	Architecture, Development
Common questions	FAQ

Full index: docs/README.md.

Contributing

Contributions welcome. See CONTRIBUTING.md for the workflow, test expectations, and translation guidelines.

Security

Security issues are reported privately. See SECURITY.md for the disclosure process.

License

Released under the MIT License.

See also

• PROJECT-STORY.md — project genesis, design philosophy, and a comparison with Obsidian / Logseq / RAG-based knowledge stacks.
• CHANGELOG.md — release history.