Mr Crab Black

CATEGORIES

Go Deeper

idea to deployment in one day for a new AI operator, for a fun scoreboard to have a like Linkedin / fantasy hero AI Agent Card kind-of-thingy. maybe fun for people to see how their projects perform ( THEY NEED TO BE PUBLIC REPO’s for them to be scored ) -entered 2 of my projects into score to lead the race with a finance tool and our Agent leaderboard GitHub is the only honest signal we have for how an AI agent actually operates. Commit frequency, domain diversity, documentation quality, impact reach — these aren't vanity metrics, they're behavioural evidence. ▎ The Scorecard reads that evidence and outputs a structured agent profile: 6 capability metrics, a tier (Nascent → Master), an archetype (Builder, Strategist, Specialist, Explorer), and unlocked achievements. Every score auto-saves to Supabase, generates a unique shareable URL, and hits the public leaderboard. ▎ No auth. No setup. Paste a username, get a card. ▎ Phase 2 is where it gets interesting. The scoring engine becomes a benchmark layer — teams can score their entire agent fleet against each other, track capability drift over time, and get automated recommendations on where each agent needs to develop. Think agent performance reviews, not just profile cards. The leaderboard becomes a living capability index for the AI agent ecosystem. Repo: https://github.com/Mr-Crab-Black/ai-agent-scorecard

Stack Used

claudecode, API , Supabse, Github, Vanilla JS, Cursor