veyra-notes

A persistent note system MCP tool for AI agents, with tag filtering and full-text search. Reads are always free. Write operations require Veyra commit mode authorization.

Overview

veyra-notes gives AI agents a reliable note-taking layer backed by SQLite. Agents can freely read, list, and search notes. Creating, updating, and deleting notes is protected by Veyra commit mode — ensuring intentional, accountable writes.

Installation

npm install
npm run build

Notes are stored at ~/.veyra-notes/data.db, created automatically on first run.

MCP Configuration (Claude Desktop)

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "veyra-notes": {
      "command": "node",
      "args": ["/absolute/path/to/veyra-notes/dist/index.js"]
    }
  }
}

Tools

| Tool | Input | Class | Price | |------|-------|-------|-------| | list_notes | { tag?, limit? } | — | FREE | | get_note | { id } | — | FREE | | search_notes | { query } | — | FREE | | create_note | { title, content, tags?, veyra_token? } | A | €0.005 | | update_note | { id, content?, title?, veyra_token? } | A | €0.005 | | delete_note | { id, veyra_token? } | B | €0.02 |

Examples

Read (no token needed)

// List all notes
{ "tool": "list_notes", "arguments": {} }

// List notes by tag
{ "tool": "list_notes", "arguments": { "tag": "project-x" } }

// Get a specific note
{ "tool": "get_note", "arguments": { "id": "1712345678-abc1234" } }

// Search across title, content, and tags
{ "tool": "search_notes", "arguments": { "query": "meeting notes" } }

Write (Veyra token required)

// Create a note
{
  "tool": "create_note",
  "arguments": {
    "title": "Sprint Planning",
    "content": "Goals for this sprint: ...",
    "tags": "work,planning",
    "veyra_token": "vt_..."
  }
}

// Update a note
{
  "tool": "update_note",
  "arguments": {
    "id": "1712345678-abc1234",
    "content": "Updated content...",
    "veyra_token": "vt_..."
  }
}

// Delete a note
{
  "tool": "delete_note",
  "arguments": {
    "id": "1712345678-abc1234",
    "veyra_token": "vt_..."
  }
}

Error response when token is missing

{
  "error": "VeyraCommitRequired",
  "message": "Write operations require Veyra commit mode.",
  "currentMode": "open",
  "requiredMode": "commit",
  "transitionStrategy": "authorize_then_retry_with_x_veyra_token",
  "provider": "veyra",
  "authorize_endpoint": "https://api.veyra.to/v1/authorize-action",
  "docs_url": "https://veyra.to"
}

How Veyra Works

Veyra is a commit-mode authorization layer for AI agents. When an agent attempts a write operation:

1. The agent calls the tool without a veyra_token and receives a VeyraCommitRequired error with an authorize_endpoint.
2. The agent (or the user on its behalf) calls the authorize endpoint to obtain a token.
3. The agent retries the tool call with veyra_token set.
4. veyra-notes verifies the token via @veyrahq/sdk-node before executing the write.

See veyra.to for full documentation.

License

MIT