Minimalistic terminal coding agent, with an emphasis on giving you control of what is sent to the model, and visibility into the model's actions.
Beautiful TUI with Tokyo Night colors, more to come soon.
We try to render diffs when the agent uses bash to edit, only for the current working directory; changes outside it aren't tracked.
npm i -g mini-coderYou can use /login and /provider to configure in the TUI. Changes are saved to your config.
Global config lives at ~/.config/mini-coder/config.json ($XDG_CONFIG_HOME/mini-coder/config.json
if set). You can also configure your provider and model there:
{
"provider": "opencode-go",
"model": "deepseek-v4.1-flash"
}Defaults for the rest:
{
"sessionsDir": "./sessions",
"systemPrompt": "",
"discoverAgentFiles": true,
"skillsDirs": [],
"tools": ["edit", "read", "bash"],
"thinkingEffort": "medium",
"customProviders": []
}provider/model— any model from thepi-aicatalog. Requires a discoverable api key from the environment (likeOPENCODE_API_KEY). When neither is set, the agent launches; sending a message reports that no model is configured.sessionsDir— where append-only session JSONL files are written. (no relative paths for now, absolute paths only).systemPrompt— the base of the system prompt. Skills and agent files, if enabled, are appended after it.discoverAgentFiles— when true, appendsAGENTS.md/CLAUDE.mdfound in~/.agentsand the working directory to the system prompt.skillsDirs— directories of skills to advertise (<dir>/<skill>/SKILL.mdwithname:anddescription:frontmatter). Paths, not contents, go to the model.tools— which ofedit,read,bashthe agent gets. Tool arguments are untrusted and validated at this boundary regardless.thinkingEffort—minimal,low,medium,high,xhigh, ormaxfor models that support reasoning.
Hosted and local models are equals. Add unlisted or self-hosted models through
customProviders — the four wire APIs come from pi-ai, and auth resolves from your
environment via envKeys (no local model needs it):
{
"customProviders": [
{
"id": "ollama",
"name": "Ollama",
"baseUrl": "http://localhost:11434/v1",
"api": "openai-completions",
"models": ["qwen3-coder:30b", "gpt-oss:20b"]
}
]
}api is openai-completions, openai-responses, anthropic-messages, or
google-generative-ai. Then point provider and model at it.