A15W v1.0.1 — Quality, Packaging & CLI Compatibility

No breaking changes. Every stage and check from v1.0.0 is preserved — this release sharpens the skills and fixes packaging.

✨ Added

npm packaging — package.json + a15w.js. Install with npm install -g a15w, run npx a15w.
Skills CLI registry — new manifest.json for the npx skill CLI.
Severity legends in every skill (FATAL/FAIL/WARN/PASS/N/A, plus SAFE/RISK/UNSAFE, STRONG/WEAK/MISSING, and impact bands).
Verification discipline sections enforcing "verify, don't assume."
Per-stage goals, ❌/✅ examples, tooling hints, and worked output examples across all 7 skills.

🔧 Changed

Restructured skills/<name>.md → skills/<name>/SKILL.md with YAML frontmatter for CLI compatibility.
CLI stats now read from manifest.json (no more hardcoded drift).
Refreshed AGENT.md and README.md.

🐛 Fixed

Broken skill paths — manifest.json and AGENT.md referenced non-existent flat files; all 7 now resolve to skills/<name>/SKILL.md.
Stat drift — corrected check count 322 → 349 everywhere.

Stats: 7 skills · 61 stages · 349 checks

Full diff: v1.0.0...v1.0.1

Release Notes

v1.0.0 — Initial Release

Release Date: May 28, 2026
Package: a15w@1.0.0
npm: https://www.npmjs.com/package/a15w
Repository: https://github.com/dev2k6/a15w

Overview

A15W (AI Agent Code Review) is a production-grade skill extension pack for AI coding agents. This initial release delivers 7 specialized pipeline skills designed to enforce quality gates on AI-generated code.

Tagline: Because "it compiled" isn't a quality metric.

Highlights

7 pipeline skills with 61 stages and 322 validation checks
Zero tolerance for hallucinations — imports, APIs, and dependencies verified against reality
Critical-first security audit — immediate abort on hardcoded secrets or injection flaws
Behavioral equivalence validation for safe refactoring operations
OpenAPI/Swagger contract enforcement for API implementations
Comprehensive test generation rejecting trivial mocks
Supply chain security audit covering CVEs, licenses, and maintenance status
Performance bottleneck identification with measurable optimization impact

Skills Arsenal

Core Skill

Skill	Stages	Description
`code_review_pipeline`	8	Macro-to-micro code review with hallucination detection. Validates business logic, dependencies, edge cases, database patterns, security, and prompt constraints.

Supporting Skills

Skill	Stages	Description
`security_audit`	10	Critical-first vulnerability detection. Hardcoded secrets, injection flaws, auth bypass, XSS, crypto weaknesses.
`test_generation`	8	Edge case coverage, error path testing, integration scenarios, property-based invariants.
`refactor_safety`	8	Behavioral equivalence validation. Same outputs, side effects, and error conditions preserved.
`dependency_audit`	8	Supply chain security. CVEs, license conflicts, abandoned packages, bundle size impact.
`api_contract_validation`	9	OpenAPI/Swagger specification fidelity. Every endpoint, status code, field, and error response.
`performance_profiling`	10	Bottleneck identification. N+1 queries, memory leaks, blocking I/O, hot paths, GC pressure.

Installation

npm (Recommended)

# View package info
npx a15w

# Install globally
npm install -g a15w

Git Clone

git clone https://github.com/dev2k6/a15w.git
cd a15w

skill CLI

npx skill add https://github.com/dev2k6/a15w.git --skill

Usage

Trigger Phrases

Phrase	Skill Activated
"review this code", "PR review", "before merging"	`code_review_pipeline`
"security audit", "vulnerability scan", "before deploy"	`security_audit`
"write tests", "generate tests", "TDD"	`test_generation`
"refactor", "is this safe", "cleanup"	`refactor_safety`
"check dependencies", "npm audit", "license compliance"	`dependency_audit`
"validate API", "OpenAPI check", "contract test"	`api_contract_validation`
"performance", "why is this slow", "optimize"	`performance_profiling`

Output Format

[PASS/FAIL/WARN] - Stage Name: Issue description & Suggested fix.

Pipeline Philosophy

Stage 1 → Stage 2 → Stage 3 → ... → Stage N
   ↓         ↓         ↓
  FAIL     FAIL      FAIL
   ↓         ↓         ↓
  STOP     STOP      STOP

Early termination on critical failures. Macro → Micro. Critical → Cosmetic.

Core Principles

Macro → Micro — Architecture before syntax
Critical → Cosmetic — Secrets before semicolons
Verify → Trust — Validate all external references
Measure → Guess — Profile before optimizing

File Structure

a15w/
├── AGENT.md              # AI agent usage protocol
├── README.md             # Human-facing documentation
├── RELEASE_NOTES_v1.0.0.md
├── manifest.json         # Skill registration
├── package.json          # npm metadata
├── bin/a15w.js           # CLI entry point
├── .gitignore
├── .gitattributes
├── LICENSE
└── skills/
    ├── manifest.json
    ├── code_review_pipeline/
    │   └── SKILL.md
    ├── security_audit/
    │   └── SKILL.md
    ├── test_generation/
    │   └── SKILL.md
    ├── refactor_safety/
    │   └── SKILL.md
    ├── dependency_audit/
    │   └── SKILL.md
    ├── api_contract_validation/
    │   └── SKILL.md
    └── performance_profiling/
        └── SKILL.md

Statistics

Metric	Value
Total Skills	7
Total Pipeline Stages	61
Total Validation Checks	322
Average Stages per Skill	8.7
Package Size (tarball)	20.1 kB
Package Size (unpacked)	57.9 kB
Total Files	13
Git Commits	14
Lines of Code/Docs	1,230+

Compatibility

Node.js: >= 16.0.0
Trae IDE: >= 1.0.0
Claude Code: >= 0.1.0
Platforms: win32, darwin, linux
Architectures: x64, arm64

Installation Methods

Method	Target	Format
npm (`npx a15w`)	CLI info	SKILL.md + manifest.json
npm (`npm install -g a15w`)	Global	`node_modules/a15w/skills/`
Git clone	Development	manifest.json at root
`npx skill add`	Skills CLI	`skills/` directory

License

MIT License — See LICENSE file for details.

Acknowledgments

Built to enforce quality standards on AI-generated code. Designed for production use. Tested against real-world hallucination patterns and security vulnerabilities.

A15W v1.0.0 — AI Agent Code Review

Stop shipping bugs. Start shipping with confidence.

Releases: dev2k6/a15w

v1.0.1