Pharos AI Agent Hackathon Adds CertiK Skill Scanner to Project Evaluation Framework

Security becomes part of AI Agent development from the start

Pharos Network has chosen CertiK’s newly launched Skill Scanner as one of the evaluation standards for its AI Agent Hackathon, placing security alongside innovation and technical execution during project assessment.

The hackathon is part of Pharos Network’s six-week global AI Agent initiative, “Create Like a PRO · Agent Carnival,” which features a 150,000 PROS prize pool worth approximately $92,500. The program is designed to support the development of AI Agents capable of operating in real-world environments rather than serving solely as demonstrations.

Participants will follow a two-stage development model. The process begins with the creation of standardized Skill modules that can be invoked by AI Agents. These Skills are then combined into fully functional Agents capable of performing tasks such as on-chain interactions, payments, collaboration, and content generation.

Projects that rank among the top performers will compete for a dedicated 50,000 PROS prize pool and may gain access to future opportunities within the broader Pharos AI ecosystem.

How Skill Scanner fits into the Pharos framework

According to Pharos, many AI Agent projects currently focus on showcasing technical concepts but lack reusable capabilities and sustainable operational models. The network argues that long-term value will come from Agents capable of autonomous asset ownership, tool utilization, and participation in on-chain economic activity.

Building such systems requires more than agent frameworks alone. Standardized Skills, trusted execution environments, and security verification mechanisms are increasingly being viewed as critical infrastructure for the next generation of AI applications.

CertiK introduced Skill Scanner in late May as a security assessment platform for AI Skill marketplaces, enterprises, and developers. The platform allows users to submit Skills through GitHub repositories, URLs, or ZIP files and evaluates them for risks such as malicious behavior, data leakage, unauthorized network access, shell execution, and file system misuse.

Unlike traditional static code analysis tools, Skill Scanner focuses on identifying risks that may emerge during runtime execution. This is particularly relevant in financial and automated decision-making environments, where vulnerabilities may only become apparent when systems are actively operating.

The integration of Skill Scanner into the hackathon’s judging framework reflects a broader shift in how AI projects are evaluated. Historically, hackathon judging has centered on innovation, user experience, and technical implementation. By adding security-focused criteria, organizers are signaling that security should be considered during development rather than after deployment.

CertiK’s Skill Scanner has already been adopted by several AI Agent ecosystems, including AI Skill marketplace Pieverse and AI Skill assetization platform FinChip.ai.

The bottom line

As AI Agents continue to evolve from experimental projects into systems capable of handling assets, executing transactions, and operating autonomously, security is becoming an increasingly important consideration. The Pharos AI Agent Hackathon’s decision to incorporate CertiK’s Skill Scanner into its evaluation process highlights a growing industry focus on building secure and reliable AI infrastructure from the outset rather than treating security as a post-launch requirement.