# Kimi Code Kimi Code is [[Moonshot AI]]'s coding-focused product built on top of the [[Kimi]] K2 model line. It is positioned as a long-horizon agentic coding surface: the use case is not single-turn code completion but multi-hour, multi-thousand-tool-call autonomous software work, similar in spirit to [[Claude Code]] and [[OpenAI Codex]] but with open-weight Kimi models underneath. ## Positioning - Long-horizon coding as the primary selling point: sustained 4,000+ tool calls over 12+ hours on top of [[Kimi K2.6]] - Agent-swarm capable: up to 300 sub-agents executing 4,000 coordinated steps (K2.6 ceiling) - Cross-language reliability: [[Rust]], Go, [[Python]] demonstrated in Moonshot's launch examples - Open-weight foundation — the underlying K2 models can be self-hosted if preferred over Moonshot's hosted product ## Use cases - Autonomous implementation of large, multi-step software tasks handed off with acceptance criteria - Converting documents (PDFs, spreadsheets) into reusable agent "Skills" - Full-stack scaffolding including front-end UI with animations, auth, and database operations - Integration as an [[Agentic Development Environment (ADE)|ADE]] backend via ACP harnesses ([[OpenClaw]], similar orchestrators) ## Ecosystem Referenced alongside other coding CLIs in ACP configurations: [[Claude Code]], [[OpenAI Codex|Codex]], OpenCode, [[Gemini CLI]], [[Pi Mono|Pi]], Kimi. Moonshot is pushing Kimi Code as the open-weight option in that lineup. ## References - Kimi platform: https://platform.kimi.ai - Kimi K2.6 announcement (first-party launch context): https://www.kimi.com/blog/kimi-k2-6 ## Related - [[Kimi CLI]] - [[Kimi]] - [[Kimi K2.6]] - [[Moonshot AI]] - [[Claude Code]] - [[OpenAI Codex]] - [[Gemini CLI]] - [[Pi Mono]] - [[OpenClaw]] - [[How Coding Agents Work]] - [[Agentic Engineering]] - [[AI Agents]] - [[AI Agent Swarms]]