armory  by Mathews-Tom

Production-grade skills and agents for serious AI coding workflows

Created 3 months ago
252 stars

Top 99.5% on SourcePulse

GitHubView on GitHub
Project Summary

Summary Armory provides a curated collection of production-grade skills, agents, and battle-tested workflows for AI coding agents. It targets developers who use AI seriously, offering repeatable, opinionated automation units that extend AI capabilities for specific task domains, enhancing productivity and reliability.

How It Works The project comprises seven package types: skills, agents, hooks, rules, commands, utilities, and presets. Each is practical, context-free, and self-contained, detailing inputs, outputs, edge cases, and failure modes. Agents orchestrate skills into multi-phase workflows, supporting routing to Claude models (Opus, Sonnet, Haiku). It integrates advanced research like EvoSkills for co-evolutionary refinement and Memento-Skills for stateful, continual learning.

Quick Start & Requirements The recommended installation is via the Skills CLI: npx skills add Mathews-Tom/armory. Alternatives include a profile installer (just install-profile <profile>), the Claude Code plugin marketplace (skills, agents, commands only), or manual cloning. Prerequisites: Node.js (for CLI), Python 3.12 (for adapter generation).

Highlighted Details

  • Extensive Package Catalog: Over 70+ curated skills and agents across 7 types, covering code review, security, research, documentation, and more.
  • Advanced AI Concepts: Integrates EvoSkills (arXiv 2604.01687) and Memento-Skills (arXiv 2603.18743) for co-evolutionary skill refinement and stateful, continual learning.
  • Cross-Platform Support: Generates adapters for Cursor, OpenAI Codex, and Gemini CLI, enabling use across diverse AI development environments.
  • Quality & Validation: Robust package evaluation system (6 dimensions) with CI/CD pipeline for automated validation.

Maintenance & Community References CONTRIBUTING.md, WANTED.md, CONTRIBUTORS.md, and ATTRIBUTIONS.md for guidelines, feature requests, contributors, and upstream acknowledgments. No direct community channels (e.g., Discord, Slack) are listed.

Licensing & Compatibility Released under the MIT license, permitting commercial use and integration with closed-source projects without significant restrictions.

Limitations & Caveats Not all package types map to all target platforms; hooks are absent on Cursor/Codex, and presets require a universal dependency resolver. The Claude Code plugin marketplace has limited support (skills, agents, commands only). Adapter generation requires Python 3.12.

Health Check
Last Commit

1 day ago

Responsiveness

Inactive

Pull Requests (30d)
8
Issues (30d)
0
Star History
18 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.