Best part of the Codex app launch is that, OpenAI has opened the whole Codex ecosystem (CLI, Web, IDE Extensions) for free ChatGPT users. And x2 usage for Plus, Pro. This is I think to gain developers' attraction from Claude Code.
One thing caught my eyes is that besides K2.5 model, Moonshot AI also launched Kimi Code (https://www.kimi.com/code), evolved from Kimi CLI. It is a terminal coding agent, I've been used it last month with Kimi subscription, it is capable agent with stable harness.
It is, Kimi Code CLI supports Zed' Agent Client Protocol (http://agentclientprotocol.com/), so it can acts as an external agent that could run in any ACP-compatible client, eg: Zed, Jetbrain, Toad CLI, Minano Notebook. Also, it supports Agent Skills. Moonshot AI developers are actively update the agent and every active. I really like their CLI.
Anecdotally, I've cancelled my Claude Code subscription after using Kimi K2.5 and Kimi CLI for the last few days. It's handled everything I've thrown at it. It is slower at the moment, but I expect that will improve.
This came as a big surprise to me last year. I remember they announced that Codex CLI is opensource, and the codex-rs [0] from TypeScript to Rust, with the entire CLI now open source. This is a big deal and very useful for anyone wanting to learn how coding agents work, especially coming from a major lab like OpenAI. I've also contributed some improvements to their CLI a while ago and have been following their releases and PRs to broaden my knowledge.
This January, I'm still working on my coding agent VT Code. Hardening execution harness, improving tool discovery mechanisms, refining context engineering and performance optimization.
Random thought: What if "AI-assisted programming" becomes "human-assisted programming" instead? We human programmers no longer play the main role in producing code, and we become the "Copilot"?
This reminds me of Amp's article last year[1]. I building my own coding agent [2]. Two goals: understand real-world agent mechanics and validate patterns I'd observed across OpenAI Codex and contemporary agents.
The core loop is straightforward: LLM + system prompt + tool calls. The differentiator is the harness, CLI, IDE extension, sandbox policies, filesystem ops (grep/sed/find). But what separates effective agents from the rest is context engineering. Anthropic and Manus has published various research articles around this topic.
After building vtcode, my takeaway: agent quality reduces to two factors, context management strategy and model capability. Architecture varies by harness, but these fundamentals remain constant.
reply