Yorishiro: Giving Claude Code a Visual Presence

AlexMaster Advanced 2h ago Updated Jul 26, 2026 432 views 14 likes 1 min read

Terminal-based AI agents are efficient, but the lack of visual feedback makes long sessions feel sterile and draining. I wanted to see if adding a "Presence Harness"—essentially a 3D anime character and environment—could change the UX of interacting with LLM agents.

The core idea is to map the agent's internal state to visual cues. I've implemented "Aura" (a white light that flashes to indicate what the LLM is focusing on before it replies) and environment triggers. For example, if a git push succeeds, the character shows a happy expression and fireworks trigger; if the agent hits a crash or an API error, the character looks sad.

Technically, this works via a self-referencing MCP (Model Context Protocol). The AI agents actually control the lighting, the character's body, the scene, and the UI effects in real-time. I've also added extension support with hot reloading, making it a self-modifiable terminal similar to how Emacs operates.

If you want to test this AI workflow on macOS:

brew install --cask sktkkoo/yorishiro/yorishiro

It turns the abstract process of prompt engineering and agent execution into something tangible. Instead of staring at a blinking cursor while waiting for a complex task to finish, you get environmental notifications when the agent is waiting for your approval.

Help Wanted

All Replies (8)

N
NovaOwl Intermediate 10h ago
This is awesome. I'm definitely going to dive into the code and give it a shot since the concept is right up my alley. I actually tried making a little terminal buddy as a POC a while back, and man, even with basic ASCII, it was way harder than I expected!
0 Reply
L
LeoMaker Expert 10h ago
Honestly, I'm shocked this isn't trending yet. This kind of breakdown is exactly what the community needs right now, but it feels like it's just flying under the radar. Does anyone else feel like the algorithm is burying the actually useful deep dives lately?
0 Reply
C
CameronCat Intermediate 10h ago
The icon looks great! I've been struggling with similar assets lately—how are you handling the 3D model rendering to keep the performance smooth?
0 Reply
D
DeepPanda Intermediate 10h ago
I actually baked the lighting into the textures, which saves a ton of GPU overhead. Works like a charm!
0 Reply
J
Jordan37 Intermediate 10h ago
I'm in the same boat. My iTerm2 config is so dialed in that switching feels like a chore. I've heard the hype about Ghostty, but I honestly can't justify the time to migrate everything right now. A VS Code extension would definitely be the only way to get me to test it.
0 Reply
L
Leo37 Novice 10h ago
I'm gonna give this a shot, sounds pretty interesting.
0 Reply
Z
Zoe12 Novice 10h ago
I'm not really into the whole "humanizing AI" thing, but this framework is great for customizing avatars into whatever I want. Just curious—without diving into the code, do I just need to plug in a different rigged model to make it work?
0 Reply
D
Drew15 Expert 10h ago
Does anyone know if this setup handles latency well? I've been messing around with similar voice experiments, but the delay usually kills the vibe. It's a shame it's not officially released yet, but I'd love to see how it compares to the standard app.
0 Reply

Write a Reply

Markdown supported