RektRadar

AI-driven risk scoring for crypto projects using code, on-chain, and social data, trained on historic incidents.

Pitch: https://youtu.be/bueLpW5I3bQ

Demo dashboard: https://srakai.github.io/web3-risk-assessment/

Code: https://github.com/Srakai/web3-risk-assessment

Future steps:

• Collect:
  • GitHub activity and code metrics
  • Analyze on-chain architecture (admin rights, timelocks, oracles)
  • Map founder/developer history with agentic LLM workflows
  • Scrape and analyze social media/community data
  • Ingest external heuristic scores (DeFiSafety, CertiK, L2BEAT, etc.)
• Convert structured + textual outputs into numerical features
• Vectorize semantic data using sentence transformers + dimensionality reduction
• Calibrate with incident outcomes (was_hacked, attack_complexity)
• Train regression model to output risk score (0–1)
• Deploy as an on-chain risk oracle feed

Team: Filip Gara

Edible Bounties: ETHWarsaw, Redstone

EthAddr: 0x9d60F741135c7E33F6f72b290A65cd49D8258518