How does cockpit talk to the AI?

The chatbot uses per project ai settings in .cockpit.json and the local bridge proxy handles AI queries.

Which port and security features are used?

Bridge runs on port 8001; writes are sandboxed to the project root with path traversal blocked. CORS is enabled for localhost during development.

How do I configure a project?

Edit the .cockpit.json in the project root to set paths, pulse rules and AI mode, then restart cockpit to apply changes.

cockpit

npx machina-cli add skill nsalvacao/nsalvacao-claude-code-plugins/cockpit --openclaw

Files (1)

SKILL.md

2.5 KB

Cockpit Skill

A browser-based dashboard that provides a visual interface for the productivity system.

Components

File	Purpose
`index.html`	Main dashboard UI
`bridge.py`	Local Python HTTP server (file I/O + AI chatbot proxy)
`cockpit.sh`	Launcher for Linux/WSL
`cockpit.bat`	Launcher for Windows
`.cockpit.json`	Per-project configuration (paths, pulse rules, AI settings)
`assets/css/styles.css`	Dashboard styles
`assets/js/app.js`	Dashboard logic

Features

Task Board — Kanban view of TASKS.md with drag-and-drop, inline editing
Memory Viewer — Browse and edit memory files (glossary, people, projects)
Project Pulse — Health check for expected files and directories
AI Chatbot — Ask questions about your project using any AI CLI or API
Spotlight Search — Ctrl+K to search across tasks, memory, and inventory
Focus Timer — Built-in Pomodoro timer
Drift Watcher — Auto-detects external file changes

Launching

The cockpit is launched via /productivity-cockpit:start, which:

Copies .cockpit.json to the project root if it doesn't exist
Starts bridge.py in the background
Opens http://localhost:8001 in the browser

AI Chatbot Configuration

The chatbot is configured per-project in .cockpit.json:

{
  "ai": {
    "mode": "cli",
    "cli": "claude",
    "args": []
  }
}

Supported modes:

CLI mode — Calls any AI CLI binary as a subprocess (claude, gemini, copilot, codex, ollama)
API mode — Direct HTTP calls to AI providers (set "mode": "api" with "provider" and env var for key)

.cockpit.json Reference

{
  "name": "My Project Cockpit",
  "version": "2.0.0",
  "paths": {
    "tasks": "TASKS.md",
    "memory": "memory",
    "output": "output"
  },
  "pulse_rules": {
    "essential_files": ["README.md"],
    "min_folders": []
  },
  "ai": {
    "mode": "cli",
    "cli": "claude",
    "args": []
  }
}

The paths section should match your project structure. The cockpit reads these paths relative to your project root.

Notes

Bridge server runs on port 8001 (auto-increments if busy)
All file writes are sandboxed to the project root (path traversal is blocked)
CORS headers are set for localhost development
The chatbot sends project context (task state, inventory) with each query

Source

git clone https://github.com/nsalvacao/nsalvacao-claude-code-plugins/blob/main/plugins/productivity-cockpit/skills/cockpit/SKILL.md

View on GitHub

Overview

Cockpit is a browser-based dashboard that visualizes your productivity system. It combines a Kanban style Task Board, a Memory Viewer, a Project Pulse health check, and an AI chatbot, all wired to a local Python bridge server for secure file I/O and AI queries. This setup makes project work more organized, observable, and chat driven within your workspace.

How This Skill Works

The UI (index.html) talks to a local bridge.py HTTP server that proxies file I/O and AI queries. A per project .cockpit.json configures paths, pulse rules, and AI settings, and the cockpit is launched via cockpit.sh or cockpit.bat. When started, it copies .cockpit.json to the project root if missing, launches bridge.py in the background, and opens http://localhost:8001 for the dashboard.

When to Use It

Starting a new project to map tasks, memory, and files.
Running regular Project Pulse checks to verify essential files and folders.
Asking the AI chatbot questions with project context for faster answers.
Using Spotlight Search to jump to tasks, memory entries, or inventory with Ctrl+K.
Developing locally with a Python bridge for fast file I/O and AI queries.

Quick Start

Step 1: Launch cockpit by running the launcher script on your OS or visit the cockpit start endpoint.
Step 2: The launcher copies the per project config, starts the bridge, and boots the UI.
Step 3: Open http://localhost:8001 and begin using the Task Board, Memory Viewer, Project Pulse, and AI Chatbot.

Best Practices

Keep a precise per project .cockpit.json with correct paths and pulse rules.
Refresh project pulse regularly to catch drift in essential files.
Leverage Spotlight Search to minimize context switching.
Choose CLI mode for low latency AI together with the local bridge, or API mode for cloud AI as needed.
Ensure the bridge is running and the dashboard is accessible at localhost:8001; all writes stay sandboxed in the project root.

Example Use Cases

A developer sets up cockpit for a new repo with TASKS.md and memory folders, then uses Project Pulse to confirm essential files.
A team member updates glossary in Memory Viewer and uses AI chatbot to query project notes.
A project lead asks the AI chatbot for sprint status and receives context enriched answers.
A user uses Ctrl+K Spotlight to jump from a task to a related memory entry while editing TASKS.md.
A local workflow uses the Python bridge to perform file I/O and AI queries without exposing keys to the browser.

Frequently Asked Questions

Add this skill to your agents