A monorepo of extensions and skills by yours truly.
TypeScript
381
423 commits
updated Sep 22, 2026
The Pi extensions and skills I use to keep long agent sessions useful without building a fake operating system around them.
Everything is published as a separate npm package. Install a bundle for the full setup, or pick only what you need. Revolutionary stuff. A table.
Pi packages run with your local permissions. You can obviously trust me, a stranger on the internet with a folder called pi-stuff, but read the source before installing it anyway.
| Package | Includes | Deliberate exclusions |
|---|---|---|
@howaboua/pi-stuff | 15 general extensions and 15 shareable skills | Browser, Codex conversion/web/image, retired standalone Code Mode, Omarchy support, and the standalone nested AGENTS loader |
@howaboua/pi-extensions | 15 general extensions | Browser, Codex conversion/web/image, retired standalone Code Mode, and the standalone nested AGENTS loader |
@howaboua/pi-skills | 15 shareable skills | Omarchy support |
pi install npm:@howaboua/pi-stuff
# or
pi install npm:@howaboua/pi-extensions
pi install npm:@howaboua/pi-skills
pi-codex-conversion is separate because it changes Pi's tool surface for GPT/Codex models. pi-codex-web-run and pi-codex-imagegen are separate because they use Codex endpoints and login. omarchy-help is separate because it targets Arch desktops configured with Omarchy.
Install pi-browser separately for logged-in browser control. The standalone pi-subdir-agents loader is also opt-in.
| Package | What it adds |
|---|---|
pi-ask | Interactive user decisions, review triage, and human handoffs |
pi-auto-trees | /marker and /end for rolling completed work into a compact branch summary |
pi-better-skills-tool | Progressive skill discovery in normal Pi, Code Mode, and Notebook Mode |
pi-browser | Logged-in browser control with local CDP sessions and SSH host routing |
pi-cache-hit-predictor | Inline prompt-cache hit predictions when switching models or reasoning levels |
pi-codex-conversion | Codex-shaped shell, patch, image inspection, and Code Mode tools for GPT/Codex models |
pi-dynamic-tools | Unmaintained standalone Code Mode, superseded by pi-codex-conversion |
pi-explore-subagents | Isolated, discovery-only shallow and deep subagents |
pi-gippity-control | Realtime voice and LAN remote control for any Pi model |
pi-gpt-switcher | /sol, /terra, and /luna commands for GPT-5.6 Codex models |
pi-codex-imagegen | Codex image generation and editing in normal Pi, Code Mode, and Notebook Mode |
pi-memories | Shutdown memory candidates in a plain Markdown inbox |
pi-pet | Animated companion miniapps for GipPity Remote |
pi-semantic-grep | Meaning-based code and docs search backed by repo-local SQLite indexes |
pi-shepherdr | Herdr-native multi-agent orchestration |
pi-smart-btw | Async side-session questions with explicit injection into the main chat |
pi-subagent-review | /review through an isolated review subagent |
pi-subdir-agents | Nested AGENTS.md context during repository discovery |
pi-unicode-charts | Terminal-native Unicode charts for Pi Markdown |
pi-vent | Batched notes about repeated workflow friction in VENT.md |
pi-codex-web-run | Codex web search and navigation in normal Pi, Code Mode, and Notebook Mode |
| Package | Use it for |
|---|---|
pi-skill-chrome-cdp | Inspecting and controlling a local Chrome-family browser through CDP |
pi-skill-code | Reviewing, changing, researching, and delivering maintained code |
pi-skill-foundations | Scoped agent guidance, clear communication, and reusable skill maintenance |
pi-skill-harness-and-agent-engineering | Diagnosing, designing, auditing, and calibrating agent harnesses |
pi-skill-omarchy-help | User-level maintenance for Arch desktops configured with Omarchy |
Pi discovers installed skills automatically and loads them when a task matches. Use /skill:<name> when you want to invoke one explicitly.
Map an unfamiliar repo, set /marker once the useful context is in place, implement one coherent change, and run /review. After triage and QA, /end carries the accepted result forward. Broad changes get a codebase-hygiene pass.
For UI work, I give the agent references first—apps, screenshots, and interface details I like—then iterate through browser inspection and screenshots. One-shotting a good frontend is mostly a party trick.
See CHANGELOG.md. Package-level changelogs remain beside packages that have them.
Individual packages include their own license files. They are MIT-licensed unless noted in the package directory.
TypeScript
68.7%
Rust
16.6%
JavaScript
10.5%
Python
3.9%
A monorepo of extensions and skills by yours truly.
TypeScript
381
423 commits
updated Sep 22, 2026
The Pi extensions and skills I use to keep long agent sessions useful without building a fake operating system around them.
Everything is published as a separate npm package. Install a bundle for the full setup, or pick only what you need. Revolutionary stuff. A table.
Pi packages run with your local permissions. You can obviously trust me, a stranger on the internet with a folder called pi-stuff, but read the source before installing it anyway.
| Package | Includes | Deliberate exclusions |
|---|---|---|
@howaboua/pi-stuff | 15 general extensions and 15 shareable skills | Browser, Codex conversion/web/image, retired standalone Code Mode, Omarchy support, and the standalone nested AGENTS loader |
@howaboua/pi-extensions | 15 general extensions | Browser, Codex conversion/web/image, retired standalone Code Mode, and the standalone nested AGENTS loader |
@howaboua/pi-skills | 15 shareable skills | Omarchy support |
pi install npm:@howaboua/pi-stuff
# or
pi install npm:@howaboua/pi-extensions
pi install npm:@howaboua/pi-skills
pi-codex-conversion is separate because it changes Pi's tool surface for GPT/Codex models. pi-codex-web-run and pi-codex-imagegen are separate because they use Codex endpoints and login. omarchy-help is separate because it targets Arch desktops configured with Omarchy.
Install pi-browser separately for logged-in browser control. The standalone pi-subdir-agents loader is also opt-in.
| Package | What it adds |
|---|---|
pi-ask | Interactive user decisions, review triage, and human handoffs |
pi-auto-trees | /marker and /end for rolling completed work into a compact branch summary |
pi-better-skills-tool | Progressive skill discovery in normal Pi, Code Mode, and Notebook Mode |
pi-browser | Logged-in browser control with local CDP sessions and SSH host routing |
pi-cache-hit-predictor | Inline prompt-cache hit predictions when switching models or reasoning levels |
pi-codex-conversion | Codex-shaped shell, patch, image inspection, and Code Mode tools for GPT/Codex models |
pi-dynamic-tools | Unmaintained standalone Code Mode, superseded by pi-codex-conversion |
pi-explore-subagents | Isolated, discovery-only shallow and deep subagents |
pi-gippity-control | Realtime voice and LAN remote control for any Pi model |
pi-gpt-switcher | /sol, /terra, and /luna commands for GPT-5.6 Codex models |
pi-codex-imagegen | Codex image generation and editing in normal Pi, Code Mode, and Notebook Mode |
pi-memories | Shutdown memory candidates in a plain Markdown inbox |
pi-pet | Animated companion miniapps for GipPity Remote |
pi-semantic-grep | Meaning-based code and docs search backed by repo-local SQLite indexes |
pi-shepherdr | Herdr-native multi-agent orchestration |
pi-smart-btw | Async side-session questions with explicit injection into the main chat |
pi-subagent-review | /review through an isolated review subagent |
pi-subdir-agents | Nested AGENTS.md context during repository discovery |
pi-unicode-charts | Terminal-native Unicode charts for Pi Markdown |
pi-vent | Batched notes about repeated workflow friction in VENT.md |
pi-codex-web-run | Codex web search and navigation in normal Pi, Code Mode, and Notebook Mode |
| Package | Use it for |
|---|---|
pi-skill-chrome-cdp | Inspecting and controlling a local Chrome-family browser through CDP |
pi-skill-code | Reviewing, changing, researching, and delivering maintained code |
pi-skill-foundations | Scoped agent guidance, clear communication, and reusable skill maintenance |
pi-skill-harness-and-agent-engineering | Diagnosing, designing, auditing, and calibrating agent harnesses |
pi-skill-omarchy-help | User-level maintenance for Arch desktops configured with Omarchy |
Pi discovers installed skills automatically and loads them when a task matches. Use /skill:<name> when you want to invoke one explicitly.
Map an unfamiliar repo, set /marker once the useful context is in place, implement one coherent change, and run /review. After triage and QA, /end carries the accepted result forward. Broad changes get a codebase-hygiene pass.
For UI work, I give the agent references first—apps, screenshots, and interface details I like—then iterate through browser inspection and screenshots. One-shotting a good frontend is mostly a party trick.
See CHANGELOG.md. Package-level changelogs remain beside packages that have them.
Individual packages include their own license files. They are MIT-licensed unless noted in the package directory.
TypeScript
68.7%
Rust
16.6%
JavaScript
10.5%
Python
3.9%