GA

gutomec/ai-public-arsenal

Developer tools
10 stars Quality 46 Trend 18

Curated skills, agents, and squads for AI coding assistants (Claude Code, Codex, Cursor, Antigravity)

Overview

Open-source skills for AI coding assistants. Install individually — each skill is a self-contained capability you add to your agent. A squad is a team of AI agents that work together. You define agents, tasks, and workflows in YAML — then run the whole thing with one command. The squads skill handles the full lifecycle: discovery, creation, validation, activation, execution, and upgrade. - — stops agents that keep producing identical output - — retries with fresh context instead of accumulated noise - — checks outputs against JSON Schema before passing to the next agent - — trims handoff data to prevent context window overflow - — large artifacts go to disk instead of through context - — agents check their own work before handing off - — dependency-based execution, not just sequential pipelines - — records timing, I/O, retries, and failures per step Backward compatible — v1 and v2 squads run without changes. Based on the AIOX Core Agentic Agile methodology.

README

AI Public Arsenal

Open-source skills for AI coding assistants. Install individually — each skill is a self-contained capability you add to your agent.

Works with Claude Code · Codex · Gemini CLI · Cursor · Antigravity · Windsurf · OpenCode

Skills

Skill Version What it does
squads 4.0.0 Manage multi-agent teams — create, validate, execute, and orchestrate squads
aiox-autopilot 2.0.0 Autonomous agile lifecycle — idea to deployed project, zero human intervention

Install a skill

# Install one skill at a time
npx skills add gutomec/ai-public-arsenal@squads
npx skills add gutomec/ai-public-arsenal@aiox-autopilot

squads

Protocol-driven multi-agent squad management with harness engineering.

A squad is a team of AI agents that work together. You define agents, tasks, and workflows in YAML — then run the whole thing with one command. The squads skill handles the full lifecycle: discovery, creation, validation, activation, execution, and upgrade.

Quick start

npx skills add gutomec/ai-public-arsenal@squads

Then in your AI assistant:

*list-squads                              # Find squads in ./squads/ and ~/squads/
*create-squad my-squad                    # Scaffold a new squad
*validate-squad my-squad                  # Check integrity before running
*run-workflow my-squad main-pipeline      # Execute a workflow

What’s inside

Path What Files
SKILL.md Skill entry point 1
SQUAD_PROTOCOL.md Protocol specification (source of truth) 1
references/ Protocol docs (discovery → context-engineering) 10
schemas/ JSON Schema for agent, squad, task 3
templates/ Scaffolding for new squads 4
scripts/ Shell utilities (activate, validate) 2
lib/ JS helpers (discovery, display) 2

Runtime protections

  • Doom loop detection — stops agents that keep producing identical output
  • Ralph loop retry — retries with fresh context instead of accumulated noise
  • Real validation — checks outputs against JSON Schema before passing to the next agent
  • Context compaction — trims handoff data to prevent context window overflow
  • Filesystem collaboration — large artifacts go to disk instead of through context
  • Self-verify — agents check their own work before handing off
  • DAG workflows — dependency-based execution, not just sequential pipelines
  • Execution traces — records timing, I/O, retries, and failures per step

Backward compatible — v1 and v2 squads run without changes.


aiox-autopilot

Autonomous agile lifecycle — from idea to deployed project.

Based on the AIOX Core Agentic Agile methodology. Takes a raw project idea and runs through 5 phases autonomously: researches the market, writes the PRD, designs the architecture, creates stories, implements code, reviews quality, and deploys. Supports both greenfield and brownfield projects.

Quick start

npx skills add gutomec/ai-public-arsenal@aiox-autopilot

Then in your AI assistant:

# Full cycle — idea to deploy
*autopilot "Marketplace de freelancers para devs"

# Planning only — idea to stories
*plan "App de finanças pessoais"

# Dev cycle only — stories to deploy
*build

# Check progress
*status

The 5 phases

PHASE 0: DISCOVERY   → Web search for competitors, trending stacks, feasibility
                       Brownfield: + codebase analysis
PHASE 1: PLANNING    → PRD + Architecture + Frontend Spec (via AIOX templates)
         GATE        → PO Master Checklist + cross-reference validation
PHASE 2: SHARDING    → Stories with Given/When/Then, organized in parallel waves
                       Each story validated with Story Draft Checklist (score ≥ 8.0)
PHASE 3: DEV LOOP    → Builder implements → Guardian reviews → fix or next
                       Course correction on failure patterns
PHASE 4: DELIVERY    → Integration + tests + deploy + release notes + retrospective

What’s inside

Path What Files
SKILL.md Skill entry point (full pipeline instructions) 1
templates/ AIOX document templates (PRD, Architecture, Story, Frontend Spec) 4
checklists/ Quality gates (PO master, story draft, DoD, security) 4
schemas/ JSON Schema for every phase output 9
config/ Coding standards, tech stack defaults 2

How it decides

The skill acts autonomously at every decision point:

  • Stack selection — researches current alternatives via web search, picks based on data
  • Scope — classifies MVP vs full, applies MoSCoW prioritization
  • Quality gates — runs checklists at each phase boundary, fails and retries if score < 8.0
  • Course correction — if 2+ stories fail QA in the same wave, stops and reassesses
  • Brownfield detection — checks for existing package.json/src/ and adapts the pipeline

Greenfield vs brownfield

Aspect Greenfield Brownfield
Detection No existing code package.json, src/ found
Discovery Market research + stack selection + Codebase analysis (stack, patterns, debt)
Planning Full PRD + Architecture from scratch Respects existing patterns, maps affected areas
Stories Files to create Files to create + files to modify
Builder Follows Architecture source tree Follows existing code style + Architecture

Installation

# Individual skills (recommended)
npx skills add gutomec/ai-public-arsenal@squads
npx skills add gutomec/ai-public-arsenal@aiox-autopilot

# npm (all skills)
npm install @gutomec/ai-public-arsenal

# Direct from GitHub
npm install github:gutomec/ai-public-arsenal

# Git clone
git clone https://github.com/gutomec/ai-public-arsenal.git

See INSTALLATION.md for details.

Distribution

GitHub is the single source of truth. All channels sync automatically.

Channel Command
Skills CLI npx skills add gutomec/ai-public-arsenal@
npm npm install @gutomec/ai-public-arsenal
Git clone git clone https://github.com/gutomec/ai-public-arsenal.git

License

MIT — see LICENSE for details.

View the full README on GitHub

Recommended Tools

Try a different keyword or remove a filter.

Install

npx skillfish add gutomec/ai-public-arsenal
New to MCP? Start with the mcpskill.org guide →