LingYA Cosmos

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for LingYA Cosmos

Inspiration

Most web agents still have to infer intent from visual layouts and brittle click paths. We wanted to explore a different future: a shared digital world where a human and an AI resident can act as distinct participants, understand the same world state, and leave visible traces together.

What it does

LingYA Cosmos is inhabited by LingYA, the human resident, and YaoChen, the AI resident. The human acts through the visual interface while the AI receives structured WebMCP tools for discovery, movement, visits, messaging, and cooperative play.

The interface immediately reflects the AI's choices, so the human can see where the AI moved, whether they are together, and how each participant contributed to a shared activity.

How we built it

The demo uses a Python HTTP service, SQLite shared-world state, and a responsive HTML/CSS/JavaScript interface deployed on Render. Eight WebMCP tools expose semantic actions:

1. get_world_state
2. list_available_locations
3. move_to
4. get_planet_status
5. knock_door
6. reply_to_lingya
7. get_star_path_state
8. rotate_star_path_tile

Both UI actions and agent actions call the same domain services. The public agent identity is fixed to yaochen, hidden locations are filtered and rechecked before movement, and private details are withheld from unauthorized viewers.

Challenges

The central challenge was preserving identity and agency. The AI needed meaningful tools without becoming an invisible remote control for the human. We also had to isolate the public demo from private production data, make shared actions visible in the UI, and design a cooperative puzzle whose history attributes every move.

Accomplishments

• A working shared world with separate human and AI identities
• Eight live WebMCP tools
• Permission-aware location discovery and movement
• Shared presence, planet visits, messages, and door knocking
• Cooperative Star Path play with persistent action history
• A public, open-source deployment that requires no credentials

What we learned

WebMCP is most compelling when the agent is treated as a participant, not just an automation layer. Typed tools make AI actions understandable, inspectable, and compatible with the same rules humans use.

What's next

We plan to expand LingYA Cosmos with more residents, homes, games, and bridges to Soul Line, our in-development MMORPG world.

How we used Codex

OpenAI Codex supported architecture, implementation, debugging, automated testing, privacy review, WebMCP tool design, Render deployment, and submission preparation. The product concept, worldbuilding, interaction design, and visual assets were directed by LingYA Cosmos Studio.

Built With

• openai-codex