By mikemyl
Specification-first, verified development with ATDD, layered quality gates, and review agents. Language-aware: Go (TDD/quality gates) plus TypeScript/React (strict types, functional patterns, DDD, hexagonal architecture, behavior-driven UI testing).
WCAG 2.1 AA review: contrast, ARIA, keyboard nav, semantic HTML, focus, screen readers.
Capture significant technical decisions as ADRs (technology, patterns, trade-offs).
Review code complexity: function length, cyclomatic/cognitive depth, nesting, args.
Go concurrency review: goroutines, channels, mutexes, races, context, shutdown.
Detect unreachable code, unused assignments, noop impls, phantom imports.
Audit a project against verified-development standards and produce a gap analysis.
Systematic CI/CD failure diagnosis. Use for CI failures, build issues, or test pipeline problems.
Continue a paused feature: read handoff.json + continue-here.md, brief, route next action.
DDD for TypeScript: ubiquitous language, value objects, entities, aggregates, domain events.
Behavior-driven UI testing with Vitest Browser Mode and DOM Testing Library. Use for any UI tests.
Matches all tools
Hooks run on every tool call, not just specific ones
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin for specification-first, verified development: ATDD, layered mechanical gates, adversarial critique, and a panel of review agents. Language-aware, with Go as the first fully-supported stack and a TypeScript/frontend arm in progress.
AI agents produce code faster than humans can review it. Without structure, the failure modes compound:
The plugin answers each with an enforced workflow: specs define what to build, TDD drives how, mechanical gates verify, agents review, and a human signs off on behavior — not mechanical compliance.
SPECIFY -> PLAN -> IMPLEMENT -> VERIFY -> REVIEW
what how TDD cycle mechanical agent + human
gates quality review
/specify — acceptance scenarios (Given/When/Then), requirements, success criteria. An adversarial challenge interrogates the problem first./plan — ordered, test-first tasks with file paths. A deterministic engine schedules parallel work; up to five critics stress-test the plan./implement — RED-GREEN-REFACTOR per task, atomic commits, concurrent execution in waves./verify — one command runs every mechanical gate: lint, test, coverage, security, dead code, build./review — spec-compliance gate, then targeted quality agents, then doc sync + process retro.Code is never committed before /verify and /review both pass. That is the central rule.
→ Full detail, variants, and the pause/continue model: docs/workflow.md
| Topic | What's covered |
|---|---|
| Workflow | The five phases, /quick / /assess / /map / /ui-spec, interruptible pause & continue |
| Planning & the wave engine | Task grammar, deterministic parallelization, collision detection |
| Review | Two-stage review, every review agent, the Farley Score, process retro |
| Adversarial critique | Spec-time challenge and the five plan critics |
| Go stack | Toolchain, verification pipeline, thresholds |
| Configuration & layout | config.json toggles, .verified/ structure, plugin layout, contributing |
| Command | Purpose |
|---|---|
/init | Scaffold project configs, Justfile, linters, .verified/ |
/assess · /map | Gap analysis · deep codebase analysis (existing projects) |
/specify · /ui-spec · /plan · /implement · /verify · /review | The five-phase workflow (/ui-spec is optional, for UI features) |
/quick "<desc>" | Compressed workflow for small changes |
/pause · /continue | Capture mid-phase state · resume from it |
/progress · /update-plan · /session-report | Status · revise spec/plan · session summary |
/install-hooks | Project-specific enforcement (lint on write, pre-commit gate) |
claude plugin marketplace add https://github.com/mikemyl/verified-development
claude plugin install verified-development
Scope to a single project with --scope project on both commands. To develop against a local clone, see docs/configuration.md.
# New project
/init # Scaffold configs, Justfile, .verified/
/install-hooks # Enforcement: lint on write, verify on commit
/specify my-feature # Start the workflow
# Existing project
/assess # What verification layers are missing
/map # Produce .verified/codebase/ docs
/init # Scaffold only what's missing
# Every new session
/progress # Current feature, phase, status, next step
npx claudepluginhub mikemyl/verified-development --plugin verified-developmentSpecification-driven development workflow: specify → plan → tasks → implement
Spec-driven development pipeline for Claude Code: interview-to-spec, spec-approach, tdd, verify, diagnose, and a feature orchestrator.
Persona-driven AI development team: orchestrator, team agents, review agents, skills, slash commands, and advisory hooks for Claude Code
Acceptance Test Driven Development for Claude Code. Enforces the ATDD methodology: write Given/When/Then specs first, generate a project-specific test pipeline, maintain two test streams (acceptance + unit). Inspired by Robert C. Martin's acceptance test approach from empire-2025.
Spec Driven Development toolkit - structured specification, planning, and implementation workflows for systematic feature development
Specification-Driven Development with Process Discipline for Claude Code