Supragent: A Standardized Skill Library for AI Agent Workflows

Supragent is a standardized skill library built for Codex and similar AI programming assistants, aiming to solve the problem of inconsistent output quality of AI assistants across different projects/teams. By encapsulating best practices for common development tasks into reusable skills, it covers multiple domains such as security management, version control, code review, and database design, helping AI assistants provide more professional and consistent responses and improve development efficiency.

With the widespread application of AI programming assistants and agents in workflows, developers face the challenge of inconsistent output quality and working methods of AI assistants across different projects. Each project has its own norms, but AI assistants lack deep understanding of specific tools and workflows. Supragent emerged as a compact and comprehensive skill library, designed to provide standardized workflow guidance for Codex, encapsulating best practices into reusable skills to make AI assistants more professional and consistent in various scenarios.

Supragent follows three core design principles:

1. Modularity and Composability: Each skill is independent, can be installed separately or combined for use, and selected on demand to avoid unnecessary burden;
2. Flexible Global and Local Deployment: Skills can be installed globally or configured locally for specific projects, balancing consistency and customization;
3. Practical Orientation: Focus on solving real development pain points, covering daily aspects such as password management, code review, Rust specifications, etc.

Supragent includes 13 skills, divided into multiple categories:

• Security and Credential Management: 1password-cli (1Password CLI key management), bitwarden-cli (Bitwarden CLI session and key retrieval);
• Version Control and Collaboration: commit (conventional commit fallback solution), pr (GitHub PR workflow guidance);
• Code Quality and Review: review-team (multi-role parallel review), modified-karpathy-guidelines (improved programming guidelines), rust-guidelines (full-process Rust guidance);
• Database and Architecture: sql-antipatterns (PostgreSQL anti-pattern audit);
• Tools and Visualization: d2lang (D2 diagram workflow), model-detector (model ID identification);
• Plugin and Extension Development: hermes-plugin-development (Hermes plugin development), stash-override-rules (Stash VPN rule writing).

Supragent is easy to install using npm skill commands:

• Install all skills globally: npx skills add caelaxie/supragent --global
• Install specific skills globally: npx skills add caelaxie/supragent --global --skill <skill-name>
• Local installation: Simply remove the --global parameter.

Supragent plays a role in various scenarios:

1. Rust Backend Project: Install rust-guidelines, sql-antipatterns, and stash-override-rules skills; AI assistants can guide writing Rust async code, avoid PostgreSQL anti-patterns, and configure VPN routes;
2. Full-Stack Project: Install 1password-cli, commit, and pr skills; AI assistants can handle keys securely, generate standardized commit messages, and create clear PRs.

Limitations: The project was created in February 2026, has a limited number of skills, few community stars, and some skills may be too general; Future Outlook: Plan to add more language-specific skills (Python, Go, etc.), cloud-native skills (Kubernetes, Terraform), special skills for test-driven development/performance optimization, and CI/CD integration skills.

Supragent encodes team knowledge and project norms into an AI-understandable form through a structured skill library, transforming AI assistants from general intelligence to domain experts while maintaining flexibility and improving output consistency. For developers using AI programming assistants like Codex, Supragent is a framework worth exploring.