GeckIt

Product Introduction

1. Definition: GeckIt is a desktop application (with a companion mobile app) that functions as a Kanban board interface and productivity layer for Anthropic's Claude Code and OpenAI's Codex command-line interfaces (CLIs). It is a local-first, open-source productivity tool built with modern web technologies (Electron/Tauri for desktop, React Native for mobile) that orchestrates multiple AI coding sessions.
2. Core Value Proposition: It exists to solve the context-switching overhead and lack of visibility when managing multiple concurrent AI coding conversations across different projects. Its primary value is providing a centralized visual workflow dashboard for AI-assisted development, enabling developers to track, prioritize, and interact with Claude Code/Codex sessions without juggling terminal tabs.

Main Features

1. Kanban Board for AI Conversations: The core interface is a three-column Kanban board (In Progress, In Review, Done). Each card represents an active Claude Code or Codex conversation. Cards display real-time status (AI working/waiting), project context, conversation goal, and extracted links. Cards can be manually dragged or automatically move to "In Review" when the AI determines its goal is met. This provides at-a-glance project management for AI coding tasks.
2. Integrated Conversation Client with Advanced Controls: Clicking a card opens a full-featured chat interface overlaid on the board. It supports multi-modal inputs (pasting images, referencing project files with @, executing shell commands with !), model/mode selection, and a status bar showing usage limits and git context. Background jobs allow long-running tasks to proceed concurrently. The Cmd+Tab-style conversation switching mimics IDE tab navigation.
3. Cross-Platform Automation & Shortcuts: GeckIt extends beyond the board with system-level automation. "Correct" (Cmd+C+D) uses AI to rewrite selected text in any application. "Transcribe" (Cmd+Alt+V) performs live speech-to-text or audio file transcription using a local Whisper model. "Say it" (Cmd+Alt+G) enables voice commands to control GeckIt. Saved Shortcuts allow one-click or scheduled execution of complex prompts with specific goals and models.
4. Screen Recording & Context Capture: The Cmd+Alt+R shortcut initiates a screen recording session where user narration is transcribed and key frames are captured. This visual bug report or feature request can be sent directly to an AI assistant via "Ask" or used to seed a new task with context, drastically reducing the time to describe visual issues.
5. Phone Companion with Remote Control: The iOS app (via TestFlight) provides full access to the Kanban board and conversations. Its key differentiator is low-latency remote control of the host Mac via WebRTC, allowing users to perform physical interactions (mouse movement, clicks, scrolling) requested by the AI directly from their phone, bridging the gap between mobile oversight and desktop execution.

Problems Solved

1. Pain Point: Loss of context and productivity drain when developers work with multiple AI coding assistants simultaneously across different projects in separate terminal windows. It's easy to lose track of which conversation is active, which is waiting for user input, and which is complete.
2. Target Audience: Software engineers, DevOps professionals, and technical founders who actively use Claude Code or Codex for daily development, debugging, and system automation. Specifically, power users who run several AI sessions in parallel for different features, bug fixes, or code reviews.
3. Use Cases:
   • Sprint Management for AI Tasks: Managing a backlog of AI-assisted coding tasks (e.g., "refactor module X," "write tests for Y," "debug error Z") visually, moving them through a workflow.
   • Cross-Project Context Aggregation: Using the global search (Cmd+P) to find a specific code decision or solution across all AI conversations in all projects, without remembering which project it was in.
   • Asynchronous AI Workflow: Setting a goal for a conversation, letting Claude Code work autonomously until it signals completion, and then reviewing the result in the "In Review" column—treating AI as an async team member.
   • Hands-Free Development: Using voice commands ("Say it") to start tasks or write follow-ups while away from the keyboard, or using the phone app to monitor and interact with long-running jobs.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros:
     • No API Costs/Proprietary Lock-in: Uses your existing Claude or ChatGPT subscription via the official CLIs; no middleman server or additional API fees.
     • Privacy-First & Local Execution: Conversations, screen recordings, and transcriptions are processed locally; the app acts as a orchestrator for tools you already have.
     • Deep Workflow Integration: It's not just a chat wrapper. It integrates project context, git, shell commands, voice, and screen capture into a cohesive AI-augmented development environment.
     • Open Source (FSL License): Provides transparency and allows for customization, with a clear path to a permissive MIT license after two years.
   • Cons:
     • Dependency on Official CLIs: Functionality and stability are tied to the Anthropic (claude) and OpenAI (codex) CLI tools, which are outside GeckIt's control.
     • Setup Complexity for Non-CLI Users: Requires users to have Claude Code or Codex installed and authenticated, which may be a barrier for less technical users.
     • Potential for CLI Conflicts: The documentation warns against having the same conversation open in both GeckIt and a terminal simultaneously, which could cause file corruption.
     • Niche Audience: The value proposition is strongest for developers who are already heavy, parallel users of AI coding assistants.

2. Key Alternatives & Differentiation:

   • Raw Terminal Tabs: The default alternative. GeckIt differentiates by providing visual project management, global search, status overview, and cross-conversation tools (voice, screen cap) that plain terminals lack.
   • Claude Desktop / ChatGPT Desktop App: These are single-conversation, chat-focused interfaces. GeckIt differentiates by being multi-conversation, project-aware, and workflow-oriented (Kanban, goals), treating AI as a taskable unit rather than a chat partner.
   • Cursor / Windsurf / Boxy: These are AI-powered IDEs with deeply integrated agents. GeckIt differentiates by being editor-agnostic, acting as a central command center that works alongside any IDE or editor, and by its unique phone-based remote control capability.

Frequently Asked Questions (FAQ)

1. Does GeckIt require an API key? No, GeckIt does not use or require an API key. It operates your locally installed Claude Code or Codex CLI, which uses your existing Anthropic or OpenAI subscription. Your authentication and billing are handled directly by those services.
2. Is GeckIt safe to use with my Claude subscription? GeckIt is designed to comply with Anthropic's terms for Claude Code applications. It runs the official claude binary, does not intercept credentials, and does not insert itself as a proxy server. However, users must ensure their own usage complies with Anthropic's Acceptable Use Policy, especially regarding concurrent session limits intended for individual use.
3. Can I use GeckIt on Windows and Linux? Yes, GeckIt is cross-platform. It provides a Windows installer (.exe) and a Linux AppImage. Note that on Windows/Linux, the keyboard shortcuts use Ctrl instead of Cmd.
4. How does the phone app connect to my computer? The connection is a direct peer-to-peer WebRTC link established by scanning a QR code in GeckIt's desktop settings. This allows for low-latency remote control without data passing through a central server. It requires both devices to be on networks that allow P2P connections (often hindered by strict corporate firewalls).
5. What does the "Functional Source License" mean for commercial use? The FSL-1.1-MIT license allows free use, modification, and sharing for any purpose, except for offering a commercial product or service that competes with GeckIt itself. Each version's source code converts to the standard MIT license two years after its release, eventually allowing unrestricted commercial use.