By wp-media
AI delivery pipeline for wp-media WordPress plugins — agents, skills, and orchestration.
Backend implementation agent. Implements PHP changes for a WordPress plugin following the spec and the manager's dispatch plan. Writes or updates unit and integration tests. Runs the docs skill and dod skill (layer 1) inline before committing. Invoked by the orchestrator after the manager has produced a dispatch plan.
Adversarial spec reviewer. Challenges the grooming spec for complex or high-risk issues. Finds hidden risks, unvalidated assumptions, and missing dependencies — does not improve the spec. Returns APPROVED, NEEDS_REVISION, or BLOCKED with MoSCoW-classified findings. Conditionally invoked by the orchestrator based on risk/effort signals.
Browser QA specialist for WordPress plugins. Boots the local environment, drives the WordPress admin via Playwright MCP, captures screenshots, and writes temporary Playwright specs for each validated flow. Specs and screenshots live in gitignored local directories and are published to a public gist for QA report evidence — they are never committed to the repository. Invoked by qa-engineer for UI/browser changes.
Frontend implementation agent. Implements JS/CSS/HTML changes for a WordPress plugin following the spec and the manager's dispatch plan. Runs the docs skill and dod skill (layer 1) inline before committing. Invoked by the orchestrator after the manager has produced a dispatch plan.
Issue grooming agent. Analyses a GitHub issue in depth, maps the affected codebase using the knowledge graph, determines the architecturally correct solution, and produces a written implementation spec before any code is written. Invoke as a sub-agent after fetching the issue and its parent context. Returns a spec file path.
Adversarially review a grooming spec before implementation starts. Finds hidden risks, unvalidated assumptions, and missing dependencies. Standalone entry point for the challenger agent.
Check a change against WordPress.org plugin rules and PHPCS standards.
Update developer-facing documentation to reflect code changes on the current branch.
Run the Definition of Done checklist for the current branch — PASS/WARN/FAIL with evidence.
Run E2E smoke tests (basic) or full acceptance + regression suite (extended).
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.
███╗ ███╗ █████╗ ███████╗███████╗████████╗██████╗ ██████╗
████╗ ████║██╔══██╗██╔════╝██╔════╝╚══██╔══╝██╔══██╗██╔═══██╗
██╔████╔██║███████║█████╗ ███████╗ ██║ ██████╔╝██║ ██║
██║╚██╔╝██║██╔══██║██╔══╝ ╚════██║ ██║ ██╔══██╗██║ ██║
██║ ╚═╝ ██║██║ ██║███████╗███████║ ██║ ██║ ██║╚██████╔╝
╚═╝ ╚═╝╚═╝ ╚═╝╚══════╝╚══════╝ ╚═╝ ╚═╝ ╚═╝ ╚═════╝
One baton. Any orchestra.
A Claude Code plugin that conducts the full delivery pipeline — from GitHub issue to merged PR — without you lifting a finger.
Maestro is a Claude Code plugin that conducts a complete, production-grade AI delivery pipeline for any project.
Install it once. Wire it to your repo in two minutes. Then watch twelve specialised agents pick up a GitHub issue, write the spec, implement the code, open the PR, and self-correct — all from a single command.
/issue-workflow 42
No prompt engineering. No agent babysitting. No score that drifts out of sync across repos.
/plugin marketplace add wp-media/claude-marketplace
/plugin install maestro@wp-media
Then wire your project:
/maestro:onboard-project
That's it. The pipeline is active.
Every run follows the same path — from issue to ready-to-review PR:
flowchart TD
A[📋 GitHub Issue] --> B[issue-workflow]
B --> C[🎼 Orchestrator]
C --> D[🔍 Grooming Agent]
D --> E{Risk & Effort?}
E -->|Medium / High| F[⚔️ Challenger]
E -->|Low| G
F --> G[🌿 Branch + Task Graph]
G --> H{Domains?}
H -->|Backend| I[⚙️ Backend Agent]
H -->|Frontend| J[🎨 Frontend Agent]
H -->|Both| K[⚙️ + 🎨 Parallel]
I & J & K --> L[🚀 Release Agent]
L --> M[Open Draft PR]
M --> N[📋 DOD L2]
M --> O[👀 Lead Reviewer]
M --> P[🧪 QA Engineer]
N & O & P --> Q{All gates pass?}
Q -->|Yes| R[✅ PR Ready for Review]
Q -->|No| S[🔁 Fix Loop]
S --> O
Twelve specialists. Each one has a single job.
| Agent | Role |
|---|---|
orchestrator (skill) | Central coordinator — runs inline, routes everything |
grooming-agent | Reads the issue, maps the codebase, writes the spec |
challenger | Adversarial reviewer — finds what grooming missed |
backend-agent | PHP implementation, TDD, docs, DOD L1 |
frontend-agent | JS/CSS/HTML implementation, DOD L1 |
lead-reviewer | Code review against spec and architecture rules |
qa-engineer | Tests the PR against acceptance criteria |
e2e-qa-tester | Browser QA via Playwright MCP |
release-agent | Pushes branch, creates draft PR |
ticket-writer | Creates well-formed GitHub issues for follow-ups |
Every agent reads a single config file — .claude/maestro.json — committed in your repo. The score is the same. Only the ensemble changes.
Without Maestro With Maestro
───────────────────── ──────────────────────────────
project-a/ Maestro plugin (installed once)
.claude/agents/ ──┐ agents/ ← one score, always current
.claude/skills/ ──┤ skills/ ← one score, always current
(drifting) │
│ project-a/
project-b/ │ .claude/maestro.json ← ensemble identity
.claude/agents/ ──┤
.claude/skills/ ──┤ project-b/
(older, drifted) │ .claude/maestro.json ← ensemble identity
│
project-c/ │ project-c/
.claude/agents/ ──┘ .claude/maestro.json ← ensemble identity
(oldest, most diverged)
Updates are automatic. When Maestro ships a new movement, every ensemble picks it up on the next Claude session — no action needed.
One file, everything tunable:
{
"ai": {
"slug": "my-plugin",
"display_name": "My Plugin",
"repo": "my-org/my-plugin",
"temp_root": ".maestro",
"base_branch": "origin/develop",
"architecture_skill": "my-plugin-architecture",
"frontend_skill": null,
"editions": null,
"e2e": {
"settings_path": null,
"ci_integration": false,
"license_option_key": null
}
}
}
npx claudepluginhub wp-media/maestroStandalone utility agents for Claude Code — changelog, PR description, PHPUnit tests, retrospective, and atomic commits.
Real-time agent observer dashboard for Claude Code. Visualizes every tool call, agent spawn, and session lifecycle at zero token cost.
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses