Kayden
Deploy once, agent runs forever.
Demo
About
The problem it solves Kayden solves the fragility of autonomous crypto agents. Most agents break when off-chain bots fail, gas runs out, or ownership/governance is unclear. Kayden gives each agent a permanent on-chain identity, deploys and tracks it through a unified dashboard, and shows verifiable activity from live contract/API data. People can use it to deploy treasury/strategy agents, monitor performance and heartbeat status, and manage agent operations with governance context. It makes autonomous workflows easier to run, easier to audit, and safer to operate. Use of AI tools and agents We used AI coding agents heavily during development: - Cursor as the primary AI IDE workflow - Claude opus 4.6 for architecture planning, code refactors, debugging, and documentation cleanup - Gemini 3.1 for cross-checking implementation details, validating edge cases, and tightening delivery under time constraints
How it's made
Challenges we ran into Biggest challenge was reliability across wallet + RPC + deployment environments at demo time. - We repeatedly hit wallet/network mismatches and transaction prompts on the wrong chain, so deploy flow got stuck. - We also hit RPC rate limits (429) when reading many agents in parallel, which made dashboard data appear broken. - On Vercel, malformed env values (trailing spaces/quotes) caused invalid address/private key failures. How I solved it: - Moved minting to a server-side /api/deploy path for stable execution. - Batched contract reads with multicall and added safer fallbacks so new agents always appear. - Added strict env sanitization/validation (trim, dequote, address/key checks) in both client and API code. - Redeployed with cleaned envs and verified all flows on live Base Sepolia.