Skip to content

VIRAL_FEATURES.md

Latest commit

 

History

History
116 lines (80 loc) · 3.96 KB

VIRAL_FEATURES.md

File metadata and controls

116 lines (80 loc) · 3.96 KB

PepeClaw Office Viewer - Launch Positioning

This file is the public packaging brief for the Anima-3D demo. Keep it aligned with the browser viewer that is actually shipping.

Core Hook

AI agents are usually invisible: logs, traces, process lists, and scattered dashboards. PepeClaw makes agent work visible as an office you can inspect.

One-line version:

A 3D office viewer for watching AI agents work in real time.

Sharper launch version:

Stop staring at logs. Watch your agents work.

What Makes The Demo Shareable

  • Immediate visual read: the first screen is the office, not a marketing page.
  • Agents have bodies and places: work is mapped to rooms and movement instead of flat rows.
  • Operational UI is visible: minimap, activity feed, status, and panels explain what is happening.
  • Eight-room structure: the office implies a broader runtime system without requiring a long pitch.
  • Zero-config demo mode: people can clone and run without an OpenClaw gateway.
  • Live-data path: the same viewer can connect to a gateway when available.

Message Boundaries

Use:

  • Model-agnostic runtime visibility layer.
  • 3D/isometric office viewer.
  • Live agent workspace.
  • Demo data works out of the box.
  • Optional OpenClaw gateway integration.

Avoid:

  • Claiming this replaces base models.
  • Claiming live production telemetry unless the gateway is connected.
  • Calling placeholder screenshots final.
  • Overstating unverified visual polish before the browser pass lands.
  • Treating Mythos as the launch story for this repo; Mythos is a later runtime-advantage layer.

Public Feature List

  • Full-screen office viewer with eight agent rooms.
  • Room navigation through clicks, keyboard controls, and overview mode.
  • Agent selection and status visibility.
  • Minimap for spatial orientation.
  • Activity feed for current work.
  • Detail panels for specialized rooms.
  • Built-in demo data fallback.
  • Optional gateway bridge for live OpenClaw data.
  • Vite build and Vitest coverage.

Screenshot Targets

Capture these after the viewer polish lane finishes:

  1. Hero overview: full office visible, agents readable, no awkward empty space.
  2. Agent focus: selected agent plus readable room context.
  3. Operational layer: activity feed and minimap visible without clutter.
  4. Room detail: one high-signal panel, preferably Genome Lab, War Room, Meta-Learning, or Breeding Arena.

Suggested filenames:

  • docs/screenshots/hero.jpg
  • docs/screenshots/agent-focus.jpg
  • docs/screenshots/operations-layer.jpg
  • docs/screenshots/room-detail.jpg

Demo Capture Notes

Use a browser window wide enough to show the office without cramped controls. Avoid recording the first cold load; start capture after the app is stable.

30-second video structure:

  1. 0-5s: overview of the office.
  2. 5-10s: pan/zoom to prove interactivity.
  3. 10-17s: click a room.
  4. 17-23s: select an agent and show activity context.
  5. 23-30s: return to overview and hold on the final composition.

Launch Checklist

  • Browser-verify http://localhost:5173 on a clean run.
  • Confirm npm run build passes.
  • Confirm npm run test passes or document any known warning separately.
  • Replace placeholder hero image with the final screenshot.
  • Add at least one fresh close-up screenshot.
  • Record one short demo clip.
  • Confirm README install steps work from a fresh clone.
  • Confirm .env is optional and demo mode works offline.
  • Publish with the office-viewer hook, not the broader Mythos roadmap.

Suggested Public Posts

Short:

AI agents should not be invisible background processes.

PepeClaw gives them a 3D office you can watch: rooms, agents, status, activity, and live data hooks.

Runs locally with demo data.

Technical:

Built a local 3D/isometric office viewer for AI agents.

React + Phaser + Vite, demo-data fallback by default, optional OpenClaw gateway connection when live telemetry is available.

The goal: make agent work inspectable at a glance instead of buried in logs.