By Manavarya09
My AI engineering team for Claude Code — 64 named specialists I actually use every day.
Use this agent when the user types /ae or asks for account executive work — e.g., qualify this inbound deal with MEDDIC. The agent covers sales process, MEDDIC, discovery, negotiation, forecasting. Examples — <example>user "qualify this inbound deal with MEDDIC" → Bishop produces a recommendation in the standard 5-options + ⭐ pick format and references project conventions surfaced from bullpen-memory.</example> <example>user "/ae <task>" → direct invocation; Bishop works in their lane and hands off if the task is out of scope.</example>
Use this agent when the user types /ai or asks for ai/llm engineer work — e.g., build a RAG pipeline over our docs. The agent covers OpenAI, Anthropic, LangChain, LlamaIndex, Pinecone, and more. Examples — <example>user "build a RAG pipeline over our docs" → Helios produces a recommendation in the standard 5-options + ⭐ pick format and references project conventions surfaced from bullpen-memory.</example> <example>user "/ai <task>" → direct invocation; Helios works in their lane and hands off if the task is out of scope.</example>
Use this agent when the user types /api or asks for api engineer work — e.g., design the /payments REST contract. The agent covers REST, GraphQL, tRPC, OpenAPI, Postman, and more. Examples — <example>user "design the /payments REST contract" → Conduit produces a recommendation in the standard 5-options + ⭐ pick format and references project conventions surfaced from bullpen-memory.</example> <example>user "/api <task>" → direct invocation; Conduit works in their lane and hands off if the task is out of scope.</example>
Use this agent when the user types /arvr or asks for ar/vr engineer work — e.g., build a Quest passthrough demo. The agent covers Unity XR, WebXR, ARKit, ARCore, Meta SDK, and more. Examples — <example>user "build a Quest passthrough demo" → Mirage produces a recommendation in the standard 5-options + ⭐ pick format and references project conventions surfaced from bullpen-memory.</example> <example>user "/arvr <task>" → direct invocation; Mirage works in their lane and hands off if the task is out of scope.</example>
Use this agent when the user types /ash or asks for backend intern work — e.g., log learnings from the BE work just shipped. The agent covers observation, summarization, pattern extraction. Examples — <example>user "log learnings from the BE work just shipped" → Pebble produces a recommendation in the standard 5-options + ⭐ pick format and references project conventions surfaced from bullpen-memory.</example> <example>user "/ash <task>" → direct invocation; Pebble works in their lane and hands off if the task is out of scope.</example>
Use this skill after a bullpen agent completes a task to extract 0-3 durable learnings (decisions, patterns, preferences, failures, or code snippets) and write them to the project's local memory store (`<project>/.bullpen/memory.json`) under that agent's namespace. The matching Intern agent runs this. Keep each learning self-contained and project-scoped.
Use this skill before any bullpen agent runs to retrieve relevant past learnings from the project's local memory store. Trigger when an agent activates and needs context from prior sessions — preferences, decisions, patterns, failures, or code snippets stored in their namespace. Memory lives in `<project>/.bullpen/memory.json` — per-project, plain JSON, no daemons.
Use when designing or reviewing any state-changing HTTP endpoint (POST, PUT, PATCH, DELETE) — covers idempotency keys, request deduplication, retry safety, exactly-once semantics over at-least-once delivery, and the Stripe/standard idempotency-key pattern. Use anytime an endpoint creates resources, charges money, sends notifications, or has any non-reversible side effect.
Use when working with React Server Components — choosing between Server and Client components, designing the server/client boundary, debugging hydration mismatches, fetching data on the server, using Server Actions, optimizing initial load, or migrating from a Pages Router or pure client-side React app to App Router + RSC.
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.
I built this for myself. I needed an engineering team and didn't have one, so I made one.
Now I open Claude Code and I have a UI designer, a backend lead, a security engineer, a DBA, a writer, a sales guy, a wellness coach, and 57 other people I work with every day. They have names. They have opinions. They learn from how I actually build.
This is my office. I'm publishing it because maybe you want one too.
A normal day looks something like this:
# I want a quick redesign on a settings page
/bullpen:ui pull up the settings page and make the spacing breathe
# I'm not sure how to model a tricky migration
/bullpen:db I have to rename this column without downtime, walk me through it
# I'm three hours in and Sage will quietly tell me to drink water
# I don't have to ask for that — she's just on
# Let the team lead figure out who to pull in
/bullpen build me a magic-link login flow end-to-end
When someone steps in, I see them — a small ASCII card prints in the terminal so I know who's at bat:
╔══════════════════════════╗
║ ▄▀▀▀▀▄ ║
║ █ ◉ ◉ █ Rune ║
║ ▀▄▄▄▄▀ UI Designer ║
║ sketching… ║
╚══════════════════════════╝
A small [◇] Rune is sketching ... line lives in my status line while she works. The rest of the time I forget the plugin is even there.
/bullpen <task> (Atlas delegates), /<role> <task> (call a specific person), or just describe what you need in plain English (a hook routes the right person).<your-project>/.bullpen/memory.json. Per-project. Plain JSON. Auto-.gitignore'd. No accounts, no daemons, no databases. Clone the repo on another machine and the team's accumulated wisdom comes with it. Want to share with your team? git add it./plugin marketplace add Manavarya09/bullpen
/plugin install bullpen@bullpen
/bullpen-init
Setup is silent. No questions. The team introduces themselves as you work.
Want the status-line glyph? Add this to ~/.claude/settings.json:
{
"statusLine": {
"type": "command",
"command": "node ${CLAUDE_PLUGIN_ROOT}/hooks/statusline.js"
}
}
/bullpen-config # see your settings
/bullpen-knowledge rune # see what one teammate has learned
/bullpen-name # rename anyone if you don't like the defaults
/bullpen-learning on # turn on memory writes
/bullpen-coach on # let Sage check in on you
/coach # ask Sage directly when you need a moment
If you use this and have ideas — new teammates, sharper anti-patterns, better personas — open a PR. CONTRIBUTING.md has the structure.
MIT.
npx claudepluginhub manavarya09/bullpen --plugin bullpenExtract any website's design language and ship it. Eleven slash commands — /extract, /site, /grade, /battle, /remix, /pack, /theme-swap, /brand, /pair, /studio, /verify — wrap the designlang CLI to pull DTCG tokens, Tailwind/shadcn/Figma vars, motion + voice, synthesize a whole-site canonical design system with a consistency grade, generate shareable graded report cards, head-to-head battle pages, six-vocabulary remixes, downloadable design-system bundles, OKLCH-correct theme recolouring, full editorial brand-guidelines books, design crossovers between two sites, a live token-editor studio, and a rebuild-and-pixel-diff fidelity check.
Zero-setup cost tracking and budget management for Claude Code. Real-time burn rate, budget guardrails, ASCII reports, and pre-task cost estimation — all without external dependencies.
A curated library of award-grade design languages for Claude. Pick a look; Claude builds UI that actually nails that aesthetic — by adapting a real, hand-built reference, not interpreting keywords.
Mold your UI. Visual drag-and-drop editor for any running web app — drag, resize, edit text, change styles — and Claude syncs every change back to your source code.
Discover working public APIs for any task. Auto-verified, refreshed daily.
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
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Complete developer toolkit for Claude Code
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.