Custos

•
3
•

Guides AI agents with official secure-coding standards, enforces compliance at commit, and generates comprehensive audit evidence for code security.

Custos functions as a secure-coding helper for AI agents, integrating official security guidance directly into the agent's context during code generation. It ensures adherence to standards like MITRE CWE, NIST 800-53, and OWASP ASVS by enforcing these guidelines at commit time, even blocking non-compliant changes. This tool provides real-time nudges, rigorous commit-time enforcement, and non-bypassable CI checks to guarantee security practices are followed. Ultimately, it streamlines the audit process by generating verifiable evidence of compliance, including mapping NIST controls to ISO 27001 Annex A, thereby proving that secure coding practices have been implemented and documented.

주요 기능

01Pre-commit and CI enforcement gates that block commits without citation receipts for security compliance.
02Real-time pre-edit security nudges for AI agents (Claude Code specific hook).
033 GitHub stars
04Automated audit evidence generation (COMPLIANCE.md) with NIST to ISO 27001 control mappings.
05In-context secure-coding guidance from official sources (CWE, NIST, OWASP ASVS) for AI agents.
06Bundled, verifiable datasets refreshed directly from official security publications.

사용 사례

01Integrate secure coding practices and compliance standards directly into AI agent code generation workflows.
02Automate the enforcement of security requirements at commit and CI stages to prevent vulnerabilities.
03Generate comprehensive audit evidence and compliance reports for codebases adhering to NIST, ISO, and OWASP standards.