veyra-tasks

A persistent task manager MCP tool for AI agents, with project grouping, priorities, and status tracking. Reads are always free. Write operations require Veyra commit mode authorization.

Overview

veyra-tasks gives AI agents a reliable task management layer backed by SQLite. Agents can freely list and inspect tasks. Creating, updating, completing, and deleting tasks is protected by Veyra commit mode — ensuring intentional, accountable writes.

Related MCP server: Cold's MCP Server - Project Management System

Installation

npm install
npm run build

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

MCP Configuration (Claude Desktop)

Add the following to your claude_desktop_config.json:

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

Tools

Status values

todo · in_progress · done

Priority values

low · medium · high · urgent

Examples

Read (no token needed)

// List all tasks
{ "tool": "list_tasks", "arguments": {} }

// List in-progress tasks for a project
{ "tool": "list_tasks", "arguments": { "status": "in_progress", "project": "website" } }

// List urgent tasks
{ "tool": "list_tasks", "arguments": { "priority": "urgent" } }

// Get a specific task
{ "tool": "get_task", "arguments": { "id": "1712345678-abc1234" } }

Write (Veyra token required)

// Create a task
{
  "tool": "create_task",
  "arguments": {
    "title": "Fix login bug",
    "description": "Users cannot log in on mobile Safari",
    "priority": "urgent",
    "project": "website",
    "due": "2025-04-20",
    "veyra_token": "vt_..."
  }
}

// Update task status
{
  "tool": "update_task",
  "arguments": {
    "id": "1712345678-abc1234",
    "status": "in_progress",
    "veyra_token": "vt_..."
  }
}

// Mark a task done
{
  "tool": "complete_task",
  "arguments": {
    "id": "1712345678-abc1234",
    "veyra_token": "vt_..."
  }
}

// Delete a task
{
  "tool": "delete_task",
  "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-tasks verifies the token via @veyrahq/sdk-node before executing the write.

See veyra.to for full documentation.

License

MIT

Hosted Pack (recommended)

Prefer the hosted pack for one-URL integration:

{
  "mcpServers": {
    "veyra": {
      "url": "https://mcp.veyra.to/sse"
    }
  }
}

One URL. 48 tools. 24 free reads. 24 protected writes.

Hosted pack: https://mcp.veyra.to/sse

Pack manifest: https://mcp.veyra.to/.well-known/veyra-pack.json

Use the hosted pack when you want the fastest MCP integration path across all Veyra tool families. Use this standalone package when you specifically want this tool on its own.

Part of the Veyra Ecosystem

Veyra is commit mode for production AI agent actions. All tools: reads free, writes require Veyra commit mode.

SDK: npm install @veyrahq/sdk-node Website: veyra.to