Aptos Labs and Almanax have announced a collaboration to develop specialized AI security agents designed to enhance security across Move-based applications.
As of November 2025, the two teams have been working on autonomous AI agents capable of detecting and remediating vulnerabilities within Move codebases. These agents are now accessible to any project within the Aptos ecosystem, spanning over 7,000 open-source Move repositories on GitHub.
Building on Aptos’s longstanding commitment to safety and formal verification, the partnership introduces autonomous AI reasoning into the development lifecycle. The new agents deeply understand Move’s resource semantics, module visibility rules, and access-control patterns, allowing them to identify vulnerabilities with high precision and produce context-rich explanations.
Developers can run full AI audits through Almanax or integrate the agents directly into DevSecOps workflows. A simple pull request command — “/almanax scan” — initiates a complete security scan, returning findings inline with reasoning traces and, when applicable, auto-remediation pull requests.
Benchmark testing using w3sa-bm-aptos, which includes real vulnerabilities from audited Aptos projects and synthetic bug variants, showed that the agents achieved performance close to human-audit parity. They demonstrated strong accuracy in access-control and resource-handling categories while significantly reducing false positives.
The same class of agents has also ranked within the top 20 on the Code4rena leaderboard, competing directly with human researchers — a milestone that highlights AI’s growing role in high-stakes security analysis.
The next phase of the collaboration focuses on expanding real-world coverage. More Aptos ecosystem projects will be onboarded to provide continuous live data, enabling further refinement of reasoning and improved precision. Newly developed features include project-level threat models, custom rule sandboxes, and automated remediation pull requests for Move vulnerabilities.
“Auditing is not a one-time effort but an ongoing process. Almanax extends this continuity by maintaining long-term vigilance,” said Andrea Cappa, Security Lead at Aptos Labs.
Through this partnership, Aptos’s safety-first foundations merge with Almanax’s advances in autonomous security agents — accelerating the path toward self-healing software systems.