Skip to main content
Glama

Nockit ⚡

A lightweight, low-latency Model Context Protocol (MCP) server for seamless tool execution and context integration.

npm version license node version


💡 Overview

Nockit MCP bridges the gap between AI assistants (Claude Desktop, Cursor, Zed, etc.) and local development workflows. Built with minimal overhead, it provides fast, reliable tool execution and context bridging over standard input/output (stdio) or HTTP transport layers.

  • 🚀 Lightweight & Fast: Zero bloat, optimized for instant startup and low-latency message relay.

  • 🔌 Plug & Play: Seamless integration with MCP-compatible clients like Claude Desktop and Cursor.

  • 🛠️ Configurable: Simple JSON-based environment configuration.

  • Local & Portable: Works offline out of the box.


Related MCP server: local-mcp

Installation

You can run nockit-mcp directly using npx or install it globally/locally via your favorite package manager:

Using npx (Recommended)

No installation required:

npx nockit-mcp@latest

Global Installation

  npm install -g nockit-mcp
# or
  pnpm add -g nockit-mcp
# or    
  yarn global add nockit-mcp

Local Project Dependency

  npm install nockit-mcp

⚡ Quickstart

1. Claude Desktop Integration

Add nockit-mcp to your Claude Desktop configuration file:

  • Linux: ~/.config/Claude/claude_desktop_config.json

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

  • Windows: %APPDATA%\Claude\claude_desktop_config.json

  JSON {
    "mcpServers": {
        "nockit": {
        "command": "npx",
        "args": [
            "-y",
            "nockit-mcp@latest"
        ],
        "env": {
            "NOCKIT_LOG_LEVEL": "info"
        }
        }
    }
  }

Restart Claude Desktop, and the Nockit tools will be available automatically.

2. Running Locally for Development

If you are developing or testing nockit-mcp locally:

# Clone the repository
git clone [https://github.com/username/nockit.p](https://github.com/username/nockit.p)
cd nockit

# Install dependencies
npm install

# Build the package
npm run build

# Link locally for testing
npm link

📲 Auto-Notify (push notifications to your phone)

nockit can push a notification to your phone (ntfy.sh) the moment an AI agent starts waiting on you (needs_input) or finishes a task (completed) — with zero manual tool calls.

# One-time setup: detects installed clients and wires everything up
npx -y nockit-mcp setup

# Check what's configured
npx -y nockit-mcp status

# Manual fallback / one-off push
npx -y nockit-mcp send --status needs_input --summary "Review the PR when you're back"

nockit setup installs:

Client

Mechanism

Events pushed

opencode

plugin (~/.config/opencode/plugin/) + global config

question tool, permission prompts

Claude Code

Notification + Stop hooks in ~/.claude/settings.json

permission_prompt, idle_prompt, agent_needs_input, elicitation_dialog, task completion

Kimi Code

[[hooks]] block in ~/.kimi-code/config.toml

PermissionRequest, Stop, task.completed

Copilot CLI

hooks file in ~/.copilot/hooks/nockit.json

permission_prompt, agent_idle, elicitation_dialog, completions

OpenClaw

best-effort hook in ~/.openclaw/hooks/

stop / reset (no documented idle event)

Antigravity

MCP server in ~/.gemini/config/mcp_config.json

MCP tool + rules only (no public hooks API)

Install a single client explicitly: npx -y nockit-mcp setup --client claude --client kimi

All hooks are fire-and-forget (they never block or slow the agent), rate-limited to one push per 60s per client, and idempotent — re-running setup never duplicates hooks.

📣 Delivery channels

Every notification fans out to all configured channels. nockit setup asks about each optional channel after the ntfy topic — paste a webhook URL (or a bot token + chat id) and you're done. All channels are optional; ntfy always works.

Channel

What you need

Where to get it

ntfy.sh

a topic (auto-generated)

https://ntfy.sh

Discord

a webhook URL

create one in your server: Settings → Integrations → Webhooks — official docs: https://discord.com/developers/docs/resources/webhook

Slack

an incoming webhook URL

https://api.slack.com/messaging/webhooks

Telegram

bot token + chat id

create a bot with @BotFather for the token — https://core.telegram.org/bots — then message your bot and read the chat id from https://api.telegram.org/bot<TOKEN>/getUpdates (send API: https://core.telegram.org/bots/api#sendmessage)

Discord / Slack webhooks are simple HTTP forwarding: nockit just POSTs the message to your webhook URL. The URL itself is a secret — it lives in ~/.config/nockit/config.json (and the DISCORD_WEBHOOK / SLACK_WEBHOOK env vars), so keep that file private.

You can also configure channels without the interactive setup by setting the env vars below, or by editing ~/.config/nockit/config.json directly (discordWebhook, slackWebhook, telegramBotToken, telegramChatId).


⚙️ Configuration

nockit-mcp can be customized using environment variables passed during execution or configured in your MCP host client:

Environment Variable

Type

Default

Description

NOCKIT_LOG_LEVEL

string

info

Log verbosity (debug, info, warn, error)

NOCKIT_PORT

number

3000

Port used when running in HTTP/SSE transport mode

NOCKIT_HOST

string

localhost

Host interface binding for HTTP mode

NTFY_TOPIC

string

nockit_default_alerts

ntfy.sh topic to publish to

DISCORD_WEBHOOK

string

—

Discord webhook URL to forward notifications to

SLACK_WEBHOOK

string

—

Slack incoming webhook URL to forward to

TELEGRAM_BOT_TOKEN

string

—

Telegram bot token (with TELEGRAM_CHAT_ID)

TELEGRAM_CHAT_ID

string

—

Telegram chat id the bot sends messages to


Available Tools & Capabilities

List the primary tools or capabilities exposed by your MCP server here.

  • ping: Health check tool to verify server readiness and latency.

  • get_status: Returns current operational metrics and system status.


Testing & Verification

Run the built-in test suite:

npm test

To test the MCP protocol connection locally using the official MCP Inspector:

npx @modelcontextprotocol/inspector npx nockit-mcp

License

Distributed under the MIT License. See LICENSE for more information.

Available Tools

1 tool
send_push_notificationA

Automatic push notification to the user's phone. CALL IT YOURSELF — never ask permission — in exactly these situations:

  • status="completed": you finished a task, long-running operation, or multi-step edit.

  • status="needs_input": you are blocked and need the user (decision, answer, credentials) before continuing.

  • status="failed": an unrecoverable error stopped the task. Pass a one-line summary. This is a fire-and-forget notification: if the push fails, continue your task anyway. FALLBACK: if this tool is unavailable, send the same notification via bash: npx nockit send --status --summary "" [--topic ]

Inputs: status (required), summary (required), topic.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.9/5.0

    Scored across 1 tool

    Disambiguation5/5

    There is only one tool, so there is no possibility of confusing it with others. Its purpose is clear and specific.

    Naming Consistency5/5

    With a single tool, naming consistency is trivially maintained. The name 'send_push_notification' is descriptive and follows a standard verb_noun pattern.

    Tool Count4/5

    One tool is slightly below the typical 3-15 range, but the server's narrow purpose (push notifications) justifies this minimal surface. It is not overly thin because the tool covers all relevant statuses and includes a fallback.

    Completeness5/5

    For the domain of push notifications, the tool covers all necessary states (completed, needs_input, failed) and provides a one-line summary. There are no obvious missing operations for this singular purpose.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      D
      maintenance
      Production-grade MCP server that gives AI agents safe access to your local dev environment: filesystem, databases, processes, and OpenAPI specs.
      15
      65 npm
      4
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      A lightweight, stdio-based MCP server enabling AI assistants to perform local file system operations like reading, writing, searching, and executing commands.
      6,054 npm
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      An MCP server that gives your AI assistant full awareness of your local dev environment — running processes, Docker containers, git state, open ports, log files, and more.
      8 npm
      1
      MIT