CertiK, the world’s largest Web3 security firm, has joined the Pharos Network AI Agent Hackathon as an industry sponsor, with its Skill Scanner tool selected as one of the event’s official evaluation standards.
Announced this Tuesday, the partnership embeds security assessment directly into the competition’s judging framework — a first for an AI Agent developer incentive program of this scale.
Security Becomes a Core Judging Criterion
The hackathon is part of Pharos Network’s “Create Like a PRO · Agent Carnival,” a six-week global initiative running through July 21 that carries a total prize pool of 150,000 PROS tokens.
Sponsored
The competition follows a two-stage model: participants first develop standardized Skill modules that can be invoked by AI Agents, then assemble those Skills into fully functional Agents capable of on-chain transactions, payments, and content generation.
Top teams will compete for a 50,000 PROS sub-pool, with winning projects gaining access to the broader Pharos AI Carnival ecosystem.
A notable feature of the competition is the integration of CertiK Skill Scanner into the judging framework.
CertiK launched Skill Scanner in late May as a security assessment platform built specifically for AI Skill marketplaces, enterprises, and developers.
The tool evaluates third-party AI Skills before execution, addressing a gap in AI Skill ecosystems where security infrastructure has not kept pace with marketplace growth.
Unlike traditional static code analysis, Skill Scanner focuses on runtime execution risks, specifically targeting vulnerabilities in financial and automated systems. Notably, its inclusion in the Pharos hackathon judging framework represents one of the first attempts to build security evaluation directly into AI Agent developer incentives.
Submissions can be reviewed via GitHub repositories, URLs, or ZIP packages, and are assessed for risks including malicious behavior, data leakage, unauthorized network access, shell execution, and file system abuse.
Pharos said the approach reflects a broader effort to move AI Agent development beyond proof-of-concept demonstrations toward sustainable, long-term operation within on-chain economies.
A Shift in How Hackathons Are Judged
Historically, hackathon judging has centered on innovation, user experience, and technical execution. The inclusion of CertiK Skill Scanner in Pharos’s evaluation criteria marks a notable shift, treating security as a prerequisite during development rather than an afterthought addressed post-launch.
Pharos noted that many AI Agent projects remain largely demonstration-oriented, lacking reusable capabilities or sustainable operating mechanisms.
The initiative is designed to push builders toward production-ready Agents capable of long-term operation, autonomous asset ownership, and on-chain economic participation.
CertiK said Skill Scanner has already seen adoption across several AI Agent ecosystems, including Pieverse, an AI Skill marketplace, and FinChip.ai, an AI Skill assetization platform.
Why This Matters
The integration of CertiK Skill Scanner into an AI Agent hackathon’s judging criteria signals a broader industry push to standardize security reviews before deployment, not after.
As AI Agents move toward managing real financial assets on-chain, embedded security evaluation at the development stage could become a baseline requirement across the sector.
Dig into DailyCoin’s popular crypto scoops today:
Zcash Falls 40% After AI-Assisted Audit Uncovers Long-Standing Bug
Will Crypto Challenge Stock Markets? Binance’s On-Chain Equity Layer
People Also Ask:
CertiK Skill Scanner is a security tool that evaluates third-party AI Skills for risks — including malicious behavior, unauthorized data access, and execution-layer vulnerabilities — before they are deployed or used by AI Agents.
It is a six-week global competition by Pharos Network offering a 150,000 PROS prize pool, challenging developers to build production-ready AI Agent Skills and fully functional on-chain Agents.
Traditionally, hackathons focus on innovation and technical execution; adding security criteria reflects the growing risk of deploying AI Agents that autonomously manage financial transactions and sensitive data.
