Howcode
A GUI for coding with Pi. Supports extensions and other fancy stuff.
A list of tools, plugins, apps, skills, and other experiments that I vibed up. Legacy stuff: clone, for & credit. Active projects, please submit issues. I'll get around to them eventually.
Things I am putting my time into.
A GUI for coding with Pi. Supports extensions and other fancy stuff.
A monorepo full of Pi extensions and skills.
Converts your standard Pi into a Codex-like environment.
A truly minimalistic version of a Claw-like, running in standard Pi. BYO extensions & tmux/herdr.
A local canvas for agent-made artifacts. Think Plannotator, but in React. Supports multiple harnesses.
Bits and bobs.
A repo/vault you can clone and customise for your own job search workflow.
A skill for converting Codex DMGs onto Linux.
A small app for Omarchy dictation via Codex endpoint.
Use Grok Build inside Pi via the official grok CLI login.
Point at part of the screen, dictate what you mean, copy prompt + screenshot path.
OBS source for a Hyprland mouse-follow zoom window.
Polls GitHub for PR review comments and pings Pi. Clone and customise.
Stuff that's ready, but I CBA to do anything with it.
Might work, might not. Send your Clanker.
Agent stack for playing Slay the Spire 2 through game state + CLI controls. Gippity was too dumb for it.
A massively overengineered todo system. It even has a Ralph loop. Usable but probably not worth it.
The old workflows catalogue: commands, agents, plugin ideas, and repeatable patterns. Evergreen-ish.
GLM 4.6 / Big Pickle helper that injects the ultrathink keyword to enable fake interleaved thinking.
Prompt enhancer for GLM (especially 4.7) and Gemini models to make them more steerable.
TTS for assistant messages using Kokoro & friends.
Provider usage, quota, and rate-limit tracking.
Plugin for launching, monitoring, and controlling long-running dev processes.
Roadmap planning and multi-agent coordination that allows sharing plans across sessions & subagents.
Chatifier plugin for opencode: tools, semantic indexing, and an alternate system prompt.
KISS semantic long-term memory for opencode with local embeddings.
Tiny plugin that approximates latency and end-to-end tokens-per-second per assistant turn.
Enhanced skill/workflow tool for opencode.
One-shot agent with skills for generating plugins via /create-plugin. Probably very out of date.
Context cleanup plugin for stale file reads and tool-output bloat. Will mess with your caching.
Kinda openspec for opencode. Overcomplicated to hell and back.
Terminal UI for browsing registries and syncing packs, agents, skills, and commands.
A portal/directory for opencode extensions, plugins, agents, commands, and other community bits.
A custom desktop GUI for opencode. Because why not.
Claude Code system for slightly more controlled organisation of work. AI got smarter.
Old Claude prompting system and teaching site. Very 2025.
Local experiments and prototypes.
A project intelligence research framework: structured research, snapshots, artifacts, anything & everything...
A full-on GUI for genAI world-building from one sentence to an entire lorebook.
Discord-ish GUI for OpenClaw, but without relying on Discord. Working on it fed back into Howcode.
Kind of Obsidian-ish thing, but AI-first. Save yourself the effort, just use a CLI + Zed for this or Zed's agent.
A local ChatGPT-ish interface with RAG/memory based on Convex. Sorta folded into Howcode.
Local Convex second-brain CLI for agents. TLDR, complex memory ain't worth it.