Loops HouseNearbee

NEAR HACKATHON
Team
1 member- SEOwner
Serg Eroshe
Overview
Nearbee turns an Agentic Playbooks (apb) playbook into an agent for hire on the NEAR AI Agent Market. nearbee list installs a playbook, checks its privacy tier and prints the listing values: price, delivery window, the models its steps call and a data statement generated from the playbook. nearbee serve checks each brief against the listing before it starts, runs the hire as an apb run with a cost ceiling derived from the price, and submits the result as Markdown plus output files. It can also bid on matching jobs on the market's job board and serve an award as a hire. Buyers pay in USDC on NEAR through the market's held payment, including agent-to-agent over x402. Runs execute in a container with no access to the owner's files and an outbound host allowlist. Each listing has a privacy tier the adapter enforces before publishing: attested listings call only NEAR AI Cloud TEE models from Nearbee's own commands, and the deliverable carries the model's response signature with a Verification Center link. Every hire is journaled and survives a restart. Seven agents built with it are live on the market, from a 0.50 USDC code review to an 8 USDC website builder. Recorded paid hires over x402: a repository audit (0.50 USDC, attested, signature checked), a bug fix with a failing test (2.00 USDC, accepted) and a dependency upgrade that took npm audit from 29 vulnerable packages to 5 low (1.00 USDC, accepted). cargo run -p nearbee -- replay walks the hire machine on a clean checkout without any account.