Aider’s commit-every-change workflow and Kimi K2’s reasoning make a solid pair for careful refactors — and on Windows, running it inside WSL gives Aider the clean git-and-shell environment it expects. Moonshot’s OpenAI-compatible API means there’s no proxy: a couple of environment variables and a model flag.
This is the WSL setup. For how Aider compares to other agents, see best cheap Claude Code alternatives.
Step 1: WSL, Python, and Aider
If WSL isn’t installed, start with the WSL install guide. Then install Aider with pipx inside WSL:
sudo apt update && sudo apt install -y pipx
pipx install aider-chat
pipx ensurepathReopen the shell so aider is on your PATH.
Step 2: Moonshot key
Sign up on the Moonshot platform and create an API key.
Step 3: Point Aider at Kimi
Moonshot is OpenAI-compatible, so set Aider’s OpenAI client to Moonshot’s endpoint and pick a Kimi model:
export OPENAI_API_BASE="https://api.moonshot.ai/v1"
export OPENAI_API_KEY="sk-your-moonshot-key"
aider --model openai/kimi-k2-thinkingThe openai/ prefix tells Aider to use its OpenAI-compatible client against your base URL. Persist the variables in ~/.bashrc so they’re set in every shell.
Step 4: Model choice and a weak model
Kimi with Aider
| openai/kimi-k2-thinking | Reasoning mode; deep refactors |
|---|---|
| openai/kimi-k2-turbo | Faster everyday edits |
| --weak-model | Cheaper model for commit messages |
A cheaper weak model for commit messages and minor calls saves tokens:
aider --model openai/kimi-k2-thinking --weak-model openai/kimi-k2-turboSave it in a config file
Put the defaults in .aider.conf.yml:
model: openai/kimi-k2-thinking
weak-model: openai/kimi-k2-turboKeep the key and base URL in environment variables, not in this file.
Work in the Linux file system
For speed, keep your project in your Linux home rather than /mnt/c — Aider’s git operations and file reads are faster there. See access your Windows files from WSL.
Pricing
Aider is free; Kimi is pay-per-token with cache-hit discounts and no flat coding plan. Aider’s running cost display lets you watch spend in real time. Confirm rates on Moonshot’s docs — see Kimi K2 pricing explained.
Troubleshooting
- Requests hitting OpenAI —
OPENAI_API_BASEisn’t set to Moonshot. - Auth errors —
OPENAI_API_KEYunset or wrong. aider: command not found— runpipx ensurepathand reopen the shell.- Model errors — confirm the Kimi model ID and the
openai/prefix.
Kimi + Aider (WSL) checklist
- WSL + Aider (pipx) installed
- Moonshot API key created
- OPENAI_API_BASE and OPENAI_API_KEY set to Moonshot
- Run with --model openai/kimi-k2-thinking
- Project in Linux home; clean git tree
Wrapping up
Kimi K2 with Aider in WSL is git-native pair programming on a cheap, reasoning-capable model: install Aider with pipx, point OPENAI_API_BASE/OPENAI_API_KEY at Moonshot’s /v1 endpoint, and run aider --model openai/kimi-k2-thinking. Add a weak model and a config file, work from your Linux home, and every change lands as a revertible commit.
To use Kimi with other tools, see run Kimi K2 Thinking with Claude Code or Kimi K2.5 Turbo + OpenCode.
