Jeongki Park
Jeongki Park is an independent developer and the creator/maintainer of Yardlet, a free, open-source native macOS app that runs real coding-agent CLIs such as Codex and Claude Code side by side in one window. Yardlet preserves each agent's own loop, sign-in, and session history while adding shared task management, account selection, approvals, review, browser, and connector capabilities.
At a Glance
- Individual software developers
- Open-source developers and technical users
- Users of Codex, Claude Code, OpenCode, Cursor Agent, and other coding-agent CLIs
- Developers managing multiple agent accounts, projects, repositories, and integrations
AI Tools by Jeongki Park
(1)Yardlet
macOS Workbench for Coding Agents
Discussions
No discussions yet
Be the first to start a discussion about Jeongki Park
Latest News
Memory and CPU of eight coding-agent apps on one Mac: Park published benchmark results comparing Yardlet 0.11.10 with seven other apps.
Yardlet 0.11.11 released with fixes for two security issues and improvements to task recovery, connector secrets, Homebrew CLI updates, diagnostics, and performance.
Yardlet 0.11.10 released with improved sign-in handling, context limits, cloud task listings, connectors, ACP agents, browser and remote-Mac support, and security changes.
Codex reviewed Claude Code's patch three times: Park documented two real regressions and one incorrect finding from cross-agent review.
Products & Services
Free, MIT-licensed native macOS application for running Codex, Claude Code, OpenCode, Cursor Agent, Antigravity CLI, Kimi Code, Grok Build, Hermes Agent, Muse Code, Pi, and other coding-agent CLIs side by side in a unified project/task window.
Market Position
Yardlet positions itself as a native, open-source multi-agent workbench rather than another coding agent: it orchestrates the CLIs users already have, retains their native sign-ins and sessions, and adds a shared UI, account/quota routing, cross-agent delegation, review, browser, and connectors. Its practical alternatives include using Codex or Claude Code directly, their dedicated apps, and other coding-agent workbenches such as Cursor Agent or OpenCode; Yardlet differentiates on native macOS implementation, MIT licensing, and support for multiple existing harnesses without replacing them.
Leadership
Founders
Jeongki Park
Independent software developer behind the GitHub account zzunkie; GitHub identifies him as Jeongki Park and links to LinkedIn at linkedin.com/in/zzunkie/. He makes and maintains Yardlet and describes it as a personal open-source project.
Executive Team
Jeongki Park
Creator and maintainer
Independent developer and GitHub user zzunkie; author of Yardlet's code, documentation, changelog, and technical blog.
Founding Story
Park says Yardlet began as his own agent workbench for working with several coding agents and accounts every day. He started it after the agent apps he was using consumed increasing amounts of Mac memory, and shared it so people without their own workbench could use one.
Business Model
Revenue Model
Yardlet is distributed as a free, open-source MIT-licensed project. The official site provides direct downloads and a Homebrew cask; no subscription, API-usage, license, or paid tier is described.
Pricing Tiers
MIT-licensed application available through GitHub Releases and Homebrew.
Target Markets
- Individual software developers
- Open-source developers and technical users
- Users of Codex, Claude Code, OpenCode, Cursor Agent, and other coding-agent CLIs
- Developers managing multiple agent accounts, projects, repositories, and integrations
- Developers who use multiple coding agents and want them in one window.
- Running work and personal accounts for Claude Code or Codex while assigning tasks to the account with available quota.
- Parallel implementation, review, testing, and deployment workflows where one agent can delegate to or consult another.
- Agent-assisted work that requires browser sessions, email/calendar/drive access, GitHub/Slack, or other MCP services.
- Reviewing diffs and pull requests alongside agent conversations and terminal work.
- Scheduling coding-agent tasks and monitoring or driving tasks from another Mac or a phone/remote client.