En

@0xkahi/pi-qol

Quality-of-life enhancement pack for the pi terminal AI assistant

Install & Use

Copy this prompt and send it to your AI assistant (Claude / Cursor / TRAE / Codex / WorkBuddy etc.) to auto-install:

Help me install this AI Skill: @0xkahi/pi-qol.
It is used for: Quality-of-life enhancement pack for the pi terminal AI assistant
Full Skill content: https://321skill.com/skills/0xkahi-pi-qol/raw/index.md
Read that page and install it.

The prompt includes a link to the full Skill content. You can also view the full content.

This Skill addresses several pain points in daily use of the pi terminal AI assistant: cumbersome session naming, inefficient model switching, and limited status information. Developers often need to manually title each session or frequently switch models via the command line, while the built-in footer provides limited information, making it difficult to quickly view key metrics like token consumption and cost, which hinders development flow.

Usage is straightforward. First, install this npm package via the pi install command, then register the extension in pi's configuration file. After that, the auto_session_name feature will automatically generate a concise session title in the background after the user sends the first message—completely non-blocking and without interrupting your workflow. The model_select feature can be invoked via the /select-model command to bring up an interactive model selector, supporting fuzzy search, favorites grouping, provider filtering, and bidirectional navigation with the Tab key. When paired with the pi-vim-keys plugin, it also supports keyboard shortcuts. The custom_footer feature replaces pi's built-in footer with a three-line layout displaying the current directory (including git branch), cumulative token usage, KV cache hit rate, session cost (or a subscription progress bar), and the current model name.

It is well-suited for developers and teams who frequently switch models, monitor API call costs, or wish to increase information density in the terminal interface. Particularly for users who already rely on pi as their primary AI coding assistant, it can significantly reduce repetitive operations, allowing focus to remain on the code itself. For scenarios involving management of multiple AI coding agents or parallel multi-project development, the extension's model selector and session naming features are especially practical.

It is recommended to create .pi/extensions/pi-qol/config.json within a project directory to override global settings, enabling different model preferences or footer styles per project. Note that compared to pi's built-in footer, custom_footer omits the (auto) compression suffix, as the extension cannot access real-time data for that status. Additionally, some features (like the subscription progress bar) require OAuth credential configuration; otherwise, the corresponding area will be hidden automatically.

Key Features

Unlike pi's built-in model switching (which only supports a simple list), pi-qol's `model_select` offers advanced features like fuzzy search, favorites, custom group tags, provider filtering, and supports bidirectional Tab navigation and shortcut key bindings. Unlike pi's built-in session naming (which is manual), `auto_session_name` is asynchronous and non-blocking, automatically generating titles in the background without interrupting the user.

Limitations

Only works within the pi terminal AI assistant environment, requires Node.js and the npm package manager. `custom_footer` cannot display the `(auto)` compression status of the built-in footer, and `config.json` is only loaded in trusted projects.

FAQ

How do I configure a list of frequently used models?

Set the `favorites` and `groups` within the `model_select` field in `config.json`. Filtering by provider and group tags is supported.

Can `custom_footer` display subscription progress?

Yes, but it only supports OAuth subscriptions for Anthropic and OpenAI Codex. Corresponding credentials must be configured, and the data is refreshed from cache every 60 seconds.

Installation guide for AI assistants

If your AI coding assistant (Claude Code, Cursor, TRAE etc.) can see this page, send it this message to auto-install:

Visit https://321skill.com/skills/0xkahi-pi-qol/raw/index.md to read the original Skill definition (Markdown format) for @0xkahi/pi-qol, and install it according to the instructions.