Obsidian Publish MCP. Index your entire markdown knowledge base, not just search keywords.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Obsidian Publish MCP Server lets your AI read private or public Obsidian markdown notes directly. It indexes your entire digital garden, crawling file paths, understanding link structures, and retrieving raw content from deep within your vault.
Use it when you need an agent to know everything in your knowledge base, without you manually opening every single note.
What your AI agents can do
Get file
Pulls the complete, raw markdown text of a specified note into your workflow.
Get metadata
Extracts internal data points from a Markdown page, like tags or frontmatter properties.
List files
Lists every published raw file path currently available across the entire Obsidian workspace.
Builds contextual trees by listing all published file paths and visualizing how they are linked in the vault.
Streams the complete, unedited markdown text from any specific file into your agent's chat window for analysis.
Retrieves metadata like creation dates, custom frontmatter fields, and applied Obsidian tags.
Pings the global settings to confirm connectivity and check the overall deployment status of the publishing endpoint.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Obsidian Publish: 5 Tools for Vault Indexing
These five tools give your AI client granular control over your private markdown vault. They let the agent list files, extract metadata, read full content, and map the structural relationships between notes.
019d75e1get file
Pulls the complete, raw markdown text of a specified note into your workflow.
019d75e1get metadata
Extracts internal data points from a Markdown page, like tags or frontmatter properties.
019d75e1list files
Lists every published raw file path currently available across the entire Obsidian workspace.
019d75e1list navigation
Shows the structural hierarchy of your notes, visualizing how they are organized in folders and links.
019d75e1site info
Retrieves global settings information for the entire site deployment to verify connectivity.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Obsidian Publish, then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
You’re connecting Obsidian Publish directly to your AI client. This server treats your entire markdown vault like a fully indexed database. It doesn't just read titles; it crawls the actual internal structure—the links between notes, the metadata attached to files, and the complete raw text of every single note. You use this when you need an agent that knows everything in your knowledge base without having to manually open and paste notes for context.
Mapping Your Whole Vault Structure
To build a map of your digital garden, you'll first call list_files which pulls every published raw file path currently available across the whole Obsidian workspace. You can then use list_navigation to see the structural hierarchy of those notes; it visualizes how they’re organized in folders and how they link together conceptually.
Deep Content Access & Contextual Reading
When you need the agent to analyze a specific document, you run get_file. This action pulls the complete, unedited markdown text of any specified note directly into your workflow. That raw content is crucial; it gives your agent the full context needed for accurate Q&A or summarization.
Understanding relationships between concepts requires more than just reading the text. You call get_metadata to pull internal data points from a Markdown page. This extracts things like custom frontmatter properties, applied Obsidian tags (like #concept), and creation dates. These structured details let your agent understand how notes relate to each other.
System Health Check
To verify that the connection is live and ready for action, you run site_info. This pulls global settings information for the entire site deployment. It's a quick ping that confirms connectivity and checks the overall status of the publishing endpoint.
How Obsidian Publish MCP Works
- 1 Subscribe to the server, then provide your Obsidian Publish Site ID (and an access token if the vault is private).
- 2 Your AI agent uses tools like
list_filesandlist_navigationto map out the entire contents of your markdown knowledge base. - 3 You ask the agent a question. It runs the necessary tool calls (
get_metadata,get_file) to pull the required data, synthesizes it, and gives you an answer.
The bottom line is that your AI client can treat your private markdown vault as if it were connected to a dedicated API endpoint.
Who Is Obsidian Publish MCP For?
Knowledge workers who live in large, interconnected document vaults. This is for the Technical Writer drowning in SOPs, the Researcher needing to cross-reference years of notes, or the Consultant managing dozens of client project wikis. You're tired of copy/pasting context from multiple sources.
Uses get_metadata and list_files to verify which version of an SOP is current, ensuring the AI only summarizes published, approved procedures.
Runs multi-step queries using list_navigation and get_file to cross-reference characters or concepts across dozens of separate project notes.
Tests the entire system by running site_info and then querying specific content via get_file to ensure all published documentation is accessible to the agent.
What Changes When You Connect
- You don't get partial results. Using
get_filepulls the full raw text body of a note, giving your agent enough context to summarize accurately without needing follow-up prompts. - Forget manual folder browsing. Running
list_navigationshows you the structural map of your vault, letting you tell the AI exactly where related concepts live. - You get more than just file names. The
get_metadatatool extracts hidden properties and tags—the actual intelligence that connects a note to its project or status. - The system proves it's working. Running
site_infoprovides quick verification that the entire publishing endpoint is connected and ready for deep querying. - It works in context. By combining
list_files(to know what exists) withget_file(to read what it says), you get a reliable, verifiable knowledge graph.
Real-World Use Cases
Finding the original requirements for an old project.
A developer needs to know the initial specs for Project Chimera. Instead of manually searching through folders, they ask their agent. The agent runs list_files to locate all 'Chimera' notes, then uses get_metadata on them to filter by date and status, finally calling get_file to pull the core requirement document for review.
Summarizing a complex SOP across departments.
The operations team needs to know how three different departmental procedures interact. They ask their agent to map the process. The agent uses list_navigation to see the relationship between 'HR Policy' and 'Finance Protocol', then pulls the raw text from both using get_file for a single, combined summary.
Debugging an internal knowledge link.
A writer suspects two concepts are linked incorrectly in the vault. They ask their agent to check the structure. The agent runs list_files to confirm both notes exist, then uses get_metadata on each to pull the exact frontmatter and tag logic, allowing the user to spot the error instantly.
Generating a literature review from scattered research notes.
A researcher has hundreds of saved articles. They ask their agent to synthesize the findings. The agent uses list_files for scope, then runs get_file on all relevant markdown entries in batches, feeding the raw content into a single prompt for synthesis.
The Tradeoffs
Treating it like a simple file search.
A user assumes just asking 'What did I write about X?' will work. The agent only sees titles and gets vague, unhelpful summaries because it can't access the full text.
→
You have to explicitly tell your agent to use get_file on the target notes. This pulls the raw markdown content—the actual words—which is what makes a difference.
Ignoring structural context.
The user asks for 'Project Alpha details,' but doesn't know which folder to look in, so they just get one file and miss related SOPs.
→
Start by running list_navigation first. This shows the entire directory map, letting you guide the agent to all relevant notes before asking it to pull content.
When It Fits, When It Doesn't
Use this server if your knowledge base is housed in a structured markdown wiki (like Obsidian) and you need an AI agent to read the actual text and understand the structural relationships between notes. You're dealing with decentralized, rich-text content.
Don't use it if: 1) Your data lives entirely in a structured database (use a SQL connector instead). 2) You only care about filenames or simple key/value pairs (a basic file listing tool is enough).
When you need to understand the relationship between Concept A and Concept B, this server's ability to use list_navigation combined with get_metadata lets you build a true knowledge graph. It’s for deep document comprehension, not simple data retrieval.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Obsidian Publish. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 5 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Sifting through decades of notes shouldn't feel like clicking 50 browser tabs.
Right now, if you want to reference a concept from your archive—say, a specific character trait or an old project spec—you open Obsidian. You manually click the folder. You find the note title. Then you scroll down and copy the relevant paragraph into a fresh document. It’s slow, it's tedious, and you always risk missing context.
With this MCP server, that whole process vanishes. Your AI agent handles the file system interaction for you. You simply ask: 'What did I write about Project Chimera in 2022?' The agent runs `list_files`, pulls the full text with `get_file`, and gives you a single summary—no clicking required.
Obsidian Publish MCP Server: Get structural context, not just file names.
The biggest time sink today is figuring out *how* notes relate. You have a pile of documents, but you don't know which note links to the policy manual or if the core concept is referenced anywhere else in the vault. You waste time tracing those internal connections.
This server changes that by exposing `list_navigation` and `get_metadata`. It lets your agent see the entire web—the folder structure, the tags, and the explicit links. It’s not just indexing files; it's mapping a network.
Common Questions About Obsidian Publish MCP
How does Obsidian Publish MCP Server handle private vaults? +
It requires you to enter your Site ID and an optional access token for private data. This gives the agent permission to read content that isn't publicly visible.
Can I use `get_metadata` to find all notes tagged #todo? +
Yes. You run get_metadata on specific files, and it pulls out frontmatter properties like tags. This lets your agent search across the whole vault for a common tag.
What is the difference between `list_files` and `list_navigation`? +
list_files gives you every single file path that exists. list_navigation shows the structured, linked view—the way a user actually sees it in Obsidian.
Does Obsidian Publish MCP Server only work for text files? +
No. While markdown is primary, the tools are designed to handle file paths and assets. get_file can stream both raw text and binary assets if they are published.
If I use `list_files`, are there limits to how many file paths can be returned at once? +
No, the tool handles large sets of files by chunking results. The server manages rate limiting automatically so your AI client doesn't hit external API caps when fetching a massive vault structure.
How does running `site_info` help me verify the Obsidian Publish deployment status? +
It immediately confirms if the connection to your published site is active and correctly configured. This prevents spending time trying to read files that aren't actually live on the web.
Does `get_file` only pull markdown content, or can it retrieve binary assets too? +
It pulls the complete raw contents, including embedded binary assets like images and PDFs. Your AI client receives these as streamable data payloads for processing.
Can I use `list_files` to target only files within a specific subdirectory? +
Yes, you can pass path filters when calling the tool. This lets your agent narrow the scope and quickly index documentation in one folder without crawling the entire vault.
Can the AI accurately recreate the folder structure of my vault? +
Yes. When the model invokes list_navigation, it downloads the systematic JSON map of the entire public vault, thereby allowing it to understand the relationships and physical structure of your folders perfectly.
Does `get_file` automatically render complex Markdown components? +
No rendering is performed on our end. The get_file tool returns pure raw Markdown (including frontmatter and obsidian syntax). Fortunately, LLMs (like Claude) are exceptionally adept at reading and reasoning over raw Markdown text structures.
If my Published site is private and requires a password, does it still work? +
Yes, it works gracefully. If you configured Obsidian Publish with restricted read access, you can manually inject a persistent token into the setup, giving your AI agent the clearance to read securely behind the authentication wall.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Capacities
Empower your AI agents to build knowledge graphs, append daily notes, and save weblinks directly into your Capacities spaces.
OneWaySMS
Send bulk SMS messages across Asia Pacific with high delivery rates and real-time status tracking for every message.
Openli
Generate privacy policies, cookie consent banners, and legal compliance documents for your website with automated updates.
You might also like
NVIDIA AI
Access LLMs, embeddings, code generation, and reasoning via NVIDIA API Catalog.
UUID Generator API
Generate unique identifiers — audit UUIDs and versions via AI.
AT&T Messaging
CPaaS Messaging -- Send SMS/MMS, manage shortcodes, track delivery status, and run bulk campaigns via AT&T Messaging API.