Submission Scoring

Each project gets a 1-5 readiness score based on how complete the submission is. This helps judges prioritize and helps builders identify what's missing.

How it works

= 3/5 readiness

Required fields carry 50%, visuals (video + images) carry 30%, recommended fields carry 20%. Projects with visuals are always shown first in the feed.

Required (50% weight)

7 fields

Project name

Must have a name

Description

Explain what you built

GitHub repo URL

Public repository required

At least 1 track

Select a competition track

Agent model specified

Which LLM powers your agent

Agent framework

e.g. LangChain, Anthropic SDK, etc.

Agent harness

e.g. Claude Code, Cursor, Codex, etc.

Recommended (20% weight)

5 fields

Problem statement

The specific challenge you're solving

Demo video URL

Strongly recommended for human judges

Live deployment URL

Judges interact with live deployments

Tools listed

Concrete tools/libraries used

Skills listed

Agent skills actually loaded

Example

A project with all required fields + video + cover image but no recommended fields scores (0.5 + 0.3) x 5 = 4/5. Add all recommended fields for a perfect 5/5. Without visuals, even a complete project maxes out at 3/5.