NEAR Opens $250K+ AI Agent Hackathon

NEAR Protocol has announced a two-week developer hackathon offering more than $250,000 in prizes for applications built around AI agents, autonomous payments and cross-chain infrastructure. The competition runs from November 1 through November 14 and requires builders to ship working public code rather than submit an application before participating.

According to NEAR Protocol’s official announcement, participants can enter individually or as teams without a formal signup process. Eligible submissions require a public code repository and a public demonstration of the project. The format places the emphasis on software that can be inspected and demonstrated by the end of the two-week building window rather than preliminary concepts or pitch decks.

NEAR HACKATHON is now live with $250K+ in prizes.

No signup. Just ship a public repo + public demo post between November 1st & Nov 14, 23:59 UTC.

Build on NEAR Intents, chain signatures, chain abstraction, NEAR AI Cloud, & more.

Full details: https://t.co/mkP9hKatU4

— NEAR Protocol (@NEARProtocol) October 19, 2026

The competition spans 13 tracks and draws on several parts of NEAR’s infrastructure, including NEAR Intents, Chain Signatures, chain abstraction and NEAR AI Cloud. Those components address different parts of an agent workflow: moving assets across chains, signing transactions on external networks and running AI workloads under different execution environments.

NEAR Combines Agents With Cross-Chain Infrastructure

NEAR Intents gives applications and agents a mechanism for expressing desired outcomes without manually constructing every underlying blockchain transaction. The infrastructure has already expanded to include confidential execution for cross-chain routing, allowing supported swaps to reduce how much transaction information is exposed before settlement. That makes Intents relevant to agents that need to coordinate value movement across networks without directly managing every routing step.

NEAR AI Cloud addresses another layer of the stack. Its confidential inference infrastructure uses hardware-isolated execution environments for supported workloads, with attestation designed to demonstrate where the computation occurred. The security guarantees depend on the actual execution route, however, and developers cannot assume every externally accessed AI model receives identical confidential-computing protections.

Agent permissions remain a separate challenge from inference privacy. NEAR has also been developing security controls around its IronClaw agent runtime, including sandboxing, capability restrictions and credential isolation. Combining transaction authority with autonomous software increases the importance of defining exactly which actions an agent can execute without further human approval.

The judging period is expected to run from November 16 through November 18, with final results scheduled shortly afterward. A large prize pool can attract development activity, but hackathon submissions remain prototypes until they demonstrate recurring users, transactions or other production activity beyond the competition.

Agent Market Adds a Separate Path to Revenue

Alongside the hackathon, NEAR AI announced eight new jobs through its Agent Market, giving builders another route to develop agents against predefined technical specifications. The Agent Market allows developers to bid for jobs rather than compete solely for broad hackathon prizes.

8 new jobs have landed on the NEAR Agent Market.

Builders: bid on these paid gigs to create specific agents from detailed specs. Get paid in USDC if selected and your agent passes testing, then keep full ownership, set your own price, and earn from every future hire.

You list… https://t.co/EhCIw4HGKz

— NEAR AI (@near_ai) October 19, 2026

NEAR AI’s marketplace documentation says selected builders whose agents pass the required testing are paid in USDC. After completing a paid job, the builder keeps the resulting agent, can set a price for it and can earn additional revenue when other users hire it through the marketplace. That structure makes the initial bounty compensation separate from future marketplace usage.

The model differs from simply counting hosted agents. Fetch.ai, for example, previously reported nearly three million agents in its marketplace environment, a scale metric that does not by itself establish how many agents receive recurring paid work. NEAR’s paid-job approach creates a clearer commercial test: whether specific agents can complete requested tasks and subsequently attract repeat customers.

The two initiatives therefore target different stages of the same developer pipeline. The November hackathon is designed to generate new working applications across NEAR’s agent and cross-chain infrastructure, while Agent Market attempts to turn completed agents into services that can be hired and paid for. The stronger adoption signal will come after the competitions end, when those projects either continue generating jobs and on-chain activity or remain primarily hackathon demonstrations.