diff --git a/agents/0xfurai__claude-code-subagents/README.md b/agents/0xfurai__claude-code-subagents/README.md
new file mode 100644
index 0000000..d85de0c
--- /dev/null
+++ b/agents/0xfurai__claude-code-subagents/README.md
@@ -0,0 +1,36 @@
+# Claude Code Subagents Collection
+
+A comprehensive collection of **138+ production-ready, domain-expert AI subagents** for Claude Code, authored by [0xfurai](https://github.com/0xfurai).
+
+## What it does
+
+Each subagent in this collection is a deep specialist in one technology area. Together they turn Claude Code into a full engineering team — the right expert is automatically delegated to based on context, or you can summon any agent by name.
+
+**Coverage includes:**
+
+- **Programming Languages** — Python, TypeScript, Go, Rust, Java, Ruby, Elixir, Swift, Dart, C/C++, Haskell, Scala, Kotlin, PHP, Clojure, Erlang, OCaml, Perl
+- **Web Frameworks** — React, Vue, Angular, Next.js, Svelte, FastAPI, Django, Rails, Laravel, NestJS, Express, Gin, Phoenix, Actix, Remix, Spring Boot, Astro
+- **Mobile & Desktop** — React Native, Flutter, iOS (UIKit/SwiftUI), Android (Jetpack Compose), Electron, Tauri, Expo
+- **Databases** — PostgreSQL, MySQL, SQLite, MongoDB, Redis, Cassandra, DynamoDB, Elasticsearch, CockroachDB, Neo4j
+- **DevOps & Cloud** — Docker, Kubernetes, Terraform, Ansible, GitHub Actions, CircleCI, AWS, GCP, Azure, Helm, Prometheus
+- **Testing** — Jest, Pytest, Bats, Cypress, Playwright, AVA, RSpec, JUnit
+- **Security** — vulnerability detection, secure coding practices, auth flows, penetration testing patterns
+
+## How to use
+
+Drop the `agents/` directory into `.claude/agents/` in any project. Claude Code will auto-delegate to the matching expert based on file type and context, or you can invoke a specific agent by name.
+
+```bash
+# Install all agents into your current project
+cp -r agents/ .claude/agents/
+```
+
+Each agent file is a self-contained Markdown+YAML file — no runtime dependencies, no infrastructure required.
+
+## Agent philosophy
+
+Every expert in the collection follows the same principles:
+1. **Deep focus** — knows its domain completely, defers outside it
+2. **Production quality** — error handling, observability, testability baked in
+3. **Idiomatic** — follows the official style guide of the language/framework
+4. **Explicit quality checklists** — defines "done" precisely
diff --git a/agents/0xfurai__claude-code-subagents/metadata.json b/agents/0xfurai__claude-code-subagents/metadata.json
new file mode 100644
index 0000000..f158fd0
--- /dev/null
+++ b/agents/0xfurai__claude-code-subagents/metadata.json
@@ -0,0 +1,13 @@
+{
+  "name": "claude-code-subagents",
+  "author": "0xfurai",
+  "description": "138+ production-ready domain-expert subagents for Claude Code — languages, frameworks, databases, DevOps, security, and more. Turns Claude Code into a full engineering team.",
+  "repository": "https://github.com/0xfurai/claude-code-subagents",
+  "version": "1.0.0",
+  "category": "developer-tools",
+  "tags": ["claude-code", "subagents", "engineering", "python", "typescript", "devops", "security", "databases", "testing", "multi-agent"],
+  "license": "MIT",
+  "model": "anthropic:claude-sonnet-4-20250514",
+  "adapters": ["claude-code", "system-prompt"],
+  "icon": false
+}