home / mcp / flint note mcp server
Agent-first note-taking system
Configuration
View docs{
"mcpServers": {
"disnet-flint-note-mcp": {
"command": "npx",
"args": [
"@flint-note/server@latest"
]
}
}
}Flint Note MCP exposes an agent-first note-taking workflow that runs as an MCP server on your machine, enabling AI clients to manage and collaborate on local Markdown notes with structured note types and metadata.
You connect an MCP client to Flint Note MCP to start creating and organizing notes through AI agents. The server reads your local markdown vaults, enforces note-type schemas, and supports cross-vault workflows without changing your active context. You can fetch multiple notes at once, filter fields to minimize data transfer, and update notes with content-hash safety to avoid conflicts.
Prerequisites: Node.js 18+ must be installed on your machine.
Install Flint Note MCP by using the provided MCP server launcher configuration for your client.
Step 1: Add Flint Note MCP to your MCP client configuration as shown here.
{
"mcpServers": {
"flint-note": {
"command": "npx",
"args": ["@flint-note/server@latest"]
}
}
}Flint Note MCP is designed to work with note types that describe how agents should handle content. Each note type has its own agent instructions and metadata schema, enabling tailored guidance for different kinds of notes like readings, meetings, projects, or daily entries.
Key workflow practices include verifying agent instructions before creating notes, using batch operations for related notes, and including content_hash when updating notes to prevent conflicts.
You can customize how agents behave by updating the agent instructions for each note type. For example you can modify prompts to emphasize deadlines, connections to related notes, or vault context.
In addition to the MCP interface, Flint Note MCP provides a direct API for programmatic access to create, search, and manage notes. The API supports creating notes, retrieving and updating content, and working with vaults and links.
Common operations include initializing the API, creating notes, fetching notes, updating content, and performing cross-vault searches.
Example operations include creating a note with type, retrieving notes, and performing a text search. You can switch between vaults to organize content by context.
Create a new note in the vault with a specified type and metadata.
Retrieve a single note by its identifier along with requested fields.
Update an existing note content or metadata, using content_hash to ensure safe revisions.
Remove a note from the vault while honoring deletion policies.
Perform full-text searches across notes with optional field filtering.
Define a new note type with its agent instructions and metadata schema.
List all available note types in the workspace.
Refine agent instructions and metadata schema for an existing note type.
List all available vaults in the current session.
Create a new vault for separate contexts such as work, personal, or research.
Switch the active vault context for subsequent operations.
Retrieve details about the currently active vault.
Perform structured searches with metadata filters, date ranges, and sorting.
Execute direct SQL-like queries for complex analytical searches.
Fetch wikilinks and references from a note to explore connections.
Find all notes that link to a given note.
Identify and report broken links within the vaults.
Create a basic note quickly with minimal inputs.
Update only the content of a note while preserving metadata.
Convenience search for quick text retrieval across notes.