Best Subagent collections in 2026: Top 21 Picks, Ranked
Ready-made specialist subagents (reviewers, testers, language experts) you copy into your agents folder or install as a plugin. Below are the 21 subagent collections we recommend in 2026, ranked by our editor score, GitHub stars, reader upvotes and how often readers click through.
Last updated October 2026 · 21 listings reviewed
1
Large plugin marketplace of specialist subagents, skills and commands for Claude Code
★ 40k · +192 this week · MIT · updated 2026-10-01
8.8/10
What it does wshobson/agents is a plugin marketplace of ready-made building blocks for agentic coding. It started as a Claude Code subagent collection and has grown into a catalogue of 94 plugins that together hold roughly 200 specialist agents, over 180 Agent Skills, more than 100 slash commands and a set of multi-agent orchestration workflows. The specialists cover architecture, individual… Read more →
Pros
- Huge range of specialists
- Organised as small installable plugins
- Actively maintained
Cons
- Easy to install too much
- Large catalogue takes time to explore
Pricing: Free · subagents plugins
2
160+ production-ready Claude Code subagents organised by category
★ 25k · +137 this week · MIT · updated 2026-09-21
8.3/10
What it does Awesome Claude Code Subagents, maintained by the team behind the VoltAgent framework, is a curated catalogue of more than 160 Claude Code subagents. Each subagent is a Markdown file with a focused role, a description that tells Claude when to delegate to it, a list of allowed tools and a preferred model. The aim is to give… Read more →
Pros
- Well organised by category
- Easy to copy single agents
Cons
- Quality varies between agents
Pricing: Free · subagents
3
Official Microsoft plugins for building Power Platform apps, pages and flows with Claude Code or Copilot
★ 947 · +19 this week · MIT · updated 2026-10-02
8.0/10
What it does Power Platform Skills is Microsoft's official plugin marketplace for building on the Power Platform with Claude Code or GitHub Copilot CLI. It is a set of plugins, one per Power Platform service, each providing skills, agents and commands. With them an agent can scaffold and deploy Power Pages code sites, build model-driven and canvas apps, generate code… Read more →
Pros
- First-party and frequently updated
- Covers pages, model/canvas/code apps, MCP tools and flows
- Installer sets up prerequisites and auto-update
Cons
- Needs PAC CLI, Azure and a Dataverse environment
- Power Pages telemetry is on by default
Pricing: Open source · power-platform power-apps power-pages microsoft plugins
4
An orchestrated dev team of Claude Code subagents with a tech-lead coordinator
★ 4.4k · MIT · updated 2025-10-30
7.6/10
What it does Awesome Claude Agents turns Claude Code into a small development team. Instead of a loose pile of specialists, it is built around orchestration: a tech-lead orchestrator breaks a feature into steps and hands each one to the most suitable specialist for your stack, while a team configurator works out which specialists your project should use in the… Read more →
Pros
- Orchestrator pattern out of the box
- Framework specialists
Pricing: Free · subagents orchestration
5
Subagents for a rapid product studio: engineering, design, marketing and ops
★ 12k · +-4 this week · updated 2025-07-28
7.5/10
What it does Contains Studio Agents is a set of Claude Code subagents modelled on the departments of a small app studio that ships in six-day sprints. Rather than focusing only on code, it gives a solo builder or tiny team stand-ins for engineering, design, marketing, product, project management, operations and testing roles. What is inside / how it works… Read more →
Pros
- Covers product, design and marketing too
- Clear role definitions
Cons
- Less active than the biggest collections
Pricing: Free · subagents product
6
175+ Codex-native .toml subagents across 13 development categories
★ 6.3k · +33 this week · MIT · updated 2026-09-21
7.5/10
What it does Awesome Codex Subagents is a curated catalogue of specialized subagent definitions written for OpenAI's Codex CLI. Each subagent is a narrow expert (an API designer, a backend developer, a security auditor, a docs researcher, and so on) that Codex can delegate part of a task to, so the main session stays focused while the specialist works in… Read more →
Pros
- Large, categorized bench of task-specific Codex experts
- Codex-native format with per-agent model routing and sandbox modes
Cons
- Definitions are community-contributed and unaudited for security or correctness
- Codex only, and requires explicit delegation
Pricing: Open source · codex subagents awesome-list code-review toml
7
A 70-agent pipeline that runs Claude Code through staged, human-gated product development
★ 102 · MIT · updated 2026-10-02
7.5/10
What it does great_cto is an orchestration layer on top of Claude Code that turns a single prompt into a full staged product-development pipeline, run by 70 specialist agents rather than one generalist session. An independent model reviews each stage before the next one builds on it, and three decisions are deliberately kept in human hands: what gets built, how… Read more →
Pros
- Structures Claude Code into a staged pipeline (brief, architecture, implementation, deployment) with an independent model reviewing each stage and three explici
- Spending caps refuse rather than silently warn, and commands like /save, /resume, /review and /trace give visibility and control over a long-running multi-agent
Cons
- 70-agent orchestration adds real token/API cost on top of whatever model you run it with, and the staged checkpoints mean it suits deliberate builds more than q
Pricing: Free · claude-code-plugin subagent-pipeline agentic-coding cost-tracking spec-driven
8
50 Claude Code subagents for authorized penetration testing, scope-gated
★ 2.3k · +31 this week · MIT · updated 2026-08-16
7.4/10
What it does pentest-ai-agents turns Claude Code into an offensive-security research assistant for authorized penetration testing. It installs around fifty subagents, each carrying deep domain knowledge in one area (recon and OSINT, web and API testing, Active Directory, cloud, containers, mobile, wireless, social engineering, credential attacks, LLM red teaming, plus defensive analysis, forensics, and reporting). You describe a task and… Read more →
Pros
- Broad specialist coverage with MITRE ATT&CK mappings and a findings DB
- Tier system, scope validation, per-command approval, and CI-enforced scope guard
Cons
- Requires written authorization and defined scope; Tier 2 runs real tools
- Claude Pro or Max needed for full use
Pricing: Open source · pentest offensive-security red-team mitre-attack subagents
9
Pi extension for foreground and background subagents with review and research agents
★ 3.8k · +54 this week · MIT · updated 2026-10-02
7.0/10
What it does pi-subagents, by Nico Bailon, is an extension for the Pi coding agent that lets it delegate work to focused child agents. The parent Pi session starts a child with its own job, then brings the result back. Children can run in the foreground and stream into the conversation, or in the background in a detached runner that… Read more →
Pros
- Ready-made scout, worker, reviewer and oracle agents
- Background runs with a live fleet inspector
- Extensive docs and a doctor command
Cons
- Only works with the Pi agent, not the apps in this directory
Pricing: Open source · pi-extension subagents code-review background-agents orchestration
10
Skills and tools that let any coding agent mod almost any PC game you own
★ 1.8k · +312 this week · MIT · updated 2026-09-30
7.0/10
What it does Universal Modder is a collection of Agent Skills, a CLI, and an MCP server that let an AI coding agent mod almost any PC game you own — reconnaissance on the game engine, reverse engineering of its file formats, AI-generated replacement assets, testing the mod in-game, and producing a showcase video of the result. What is inside… Read more →
Pros
- Covers the full modding loop end to end: engine recon, reverse engineering, AI-generated art/3D/audio via fal, asset conversion, in-game testing, and showcase v
- Works with any agent that reads Agent Skills or AGENTS.md, not locked to one vendor, and installs through existing plugin marketplaces
Cons
- Niche use case outside typical app/web development; reverse-engineering game files can run into a game's EULA or anti-cheat terms depending on the title, which
Pricing: Free · agent-skills game-modding reverse-engineering mcp claude-code-plugin
11
Plugin marketplace of skills, hooks, agents and MCP configs for four coding agents
★ 1.2k · +6 this week · Apache-2.0 · updated 2026-10-02
7.0/10
What it does Claude Codex Settings, maintained by Fatih Cagatay Akyon, is a plugin marketplace of skills, hooks, subagents, commands and MCP configurations for AI coding tools. It began as a personal setup and is now split into many small plugins, so you install only the pieces you want instead of copying a whole configuration folder. A shared CLAUDE.md with… Read more →
Pros
- Install only the plugins you need instead of a whole config
- Same review and writing helpers across Claude Code, Codex, Gemini and Cursor
- Updated very frequently
Cons
- Quality and relevance vary across a very broad plugin list
- Several plugins install hooks that run on every write or commit
Pricing: Open source · claude-code-plugin hooks skills codex mcp-config
12
Maps and documents codebases of any size with parallel Claude Code subagents
★ 739 · +7 this week · MIT · updated 2026-05-13
7.0/10
What it does Cartographer maps and documents a codebase so that coding agents can navigate it without reading every file. The classic mode is a Claude Code plugin that orchestrates parallel Sonnet subagents: it scans the file tree with token counts, plans how to split the work across budgets, spawns subagents that each analyse a slice, then synthesises their reports… Read more →
Pros
- Generates a detailed architecture map and only re-analyses changed modules
- Adds a queryable SQLite code graph and MCP wrapper in v2
Cons
- Spawns Sonnet subagents and can use significant tokens
- Claude Code only; no separate LICENSE file despite MIT claim
Pricing: Open source · codebase-mapping documentation subagents code-graph mcp
13
Self-learning skill layer for Claude Code that distills, merges and prunes skills from your real sessions
★ 7.2k · +2.6k this week · MIT · updated 2026-10-02
6.5/10
What it does Autoharness is a Claude Code plugin that writes and maintains skills for you, based on the work you actually do. It watches your sessions through hooks and counts tool calls. Once a session has done enough real work, a background reflection looks at what happened and decides whether a new skill is worth keeping. It can also… Read more →
Pros
- Learns and maintains skills automatically from real work
- Touches only its own skills, leaving yours untouched
- Keeps a ledger and evidence for every change
Cons
- Background reflection spawns extra sessions and uses more tokens
- Claude Code only, needs python3 on Linux or macOS
Pricing: Open source · agent-skills claude-code plugin self-learning automation
14
Plugin marketplace and discovery site for Claude Code agents, commands, hooks and skills
★ 3.6k · +33 this week · MIT · updated 2026-09-30
6.5/10
What it does Build with Claude is a Claude Code plugin marketplace with a companion website at buildwithclaude.com. The repository holds a large set of subagents, slash commands, hooks and skills, grouped into plugins by topic. You can install one plugin at a time or pull in everything of one type at once. The website does two jobs. You can… Read more →
Pros
- Broad set of agents, commands, hooks and skills
- Website indexes the wider community ecosystem too
- Install one plugin or whole bundles
Cons
- Quality varies across many contributors
- Some agents only loosely relate to everyday coding
Pricing: Open source · claude-code plugins subagents commands marketplace
15
Starter Claude Code config with a CLAUDE.md memory-bank system, hooks, subagents and commands
★ 2.6k · +6 this week · MIT · updated 2026-09-28
6.5/10
What it does My Claude Code Setup is a starter configuration for Claude Code projects, shared by the author of the Centmin Mod project. Its centrepiece is a "memory bank" system built on CLAUDE.md: a set of memory files and instructions designed to keep context across many chat sessions, so the agent retains project knowledge instead of relearning it each… Read more →
Pros
- CLAUDE.md memory-bank templates aimed at retaining context across sessions
- Bundles statusline, worktree scripts, subagents and slash-command groups
- Actively updated to follow Anthropic guidance
Cons
- No installer; you copy and trim files by hand
- Contains affiliate/discount links unrelated to the tooling
Pricing: Open source · claude-code claude-md memory-bank starter-template subagents
16
Curated skills, subagents and multi-provider config templates for Claude Code
★ 1.7k · +4 this week · MIT · updated 2026-09-27
6.5/10
What it does Claude Code Settings is one developer's published Claude Code setup, packaged so other people can reuse it. The main plugin bundles a set of skills and subagents for research, code review, issue fixing, image generation and session handoffs. The repository also has settings.json templates for running Claude Code through other model providers and gateways. The README is… Read more →
Pros
- Practical GitHub, research and review skills
- Config templates for many model providers
- Skills also installable into other agents
Cons
- README is Chinese only
- Manual install overwrites your ~/.claude folder
Pricing: Open source · claude-code skills subagents deep-research config-templates
17
Senior-architect development harness for Pi with subagents, guardrails and optional spec-driven flow
★ 1.2k · +84 this week · MIT · updated 2026-10-02
6.5/10
What it does Gentle Shell, published on npm as gentle-pi, is a package for the Pi coding agent. It turns Pi into a structured development workspace. It adds a persona called el Gentleman that clarifies scope before building. It sends focused subagents off to map, implement or verify code while one parent session stays in charge. It also adds workspace… Read more →
Pros
- Structured workspace with change tracking and agent orchestration
- Native review scoped to one frozen change
- Guardrails around destructive operations
Cons
- Only works with the Pi agent, not the directory's apps
- Large surface area with a real learning curve
Pricing: Open source · pi harness subagents code-review tdd
18
Plugin marketplace of Power BI and Fabric skills, agents and hooks for Claude Code and Copilot
★ 957 · +17 this week · GPL-3.0 · updated 2026-10-02
6.5/10
What it does Power BI Agentic Development is a plugin marketplace that teaches coding agents how to work with Power BI and Microsoft Fabric. It packages domain knowledge, subagents and validation hooks so an agent can help build and maintain semantic models, reports, paginated reports, custom visuals and ETL, along with the workspaces and deployment pipelines around them. Instead of… Read more →
Pros
- Focused plugins for semantic models, reports, PBIP, Tabular Editor and Fabric
- Deterministic hooks validate TMDL, PBIR, DAX and referential integrity
- Works in both Claude Code and Copilot CLI from one manifest
Cons
- In a deliberate breaking-transition version range; pin an older release if you need stability
- Only useful within the Microsoft BI stack
Pricing: Open source · power-bi microsoft-fabric plugin-marketplace hooks review-agents
19
Curated list of tools, subagents, skills, plugins and MCP servers for OpenAI Codex CLI
★ 540 · +9 this week · CC0-1.0 · updated 2026-09-06
6.5/10
What it does Awesome Codex CLI is a reading list for OpenAI's terminal coding agent. It gathers official documentation, community subagents, skills, plugins, hooks, MCP servers and companion tools into one organised page. Each entry has a short note on what it is and, often, why it is worth a look. It is a discovery resource, not something you install.… Read more →
Pros
- Groups a fragmented ecosystem by feature category with short notes
- Covers official docs, subagents, skills, plugins, MCP servers and tooling
- Live star badges help gauge popularity
Cons
- A discovery list, nothing to install; you must vet each linked project yourself
- Open PR-based list, so curation depth depends on the maintainer
Pricing: Free · awesome-list codex-cli subagents mcp-servers discovery
20
A Codex skill that splits complex tasks across Explorer, Executor, Reviewer and Advisor sub-agents
★ 213 · +47 this week · MIT · updated 2026-10-01
6.5/10
What it does Codex Team Mode is a Codex Skill that turns a single Codex session into a small coordinated team. A primary agent handles task decomposition and final validation while four specialized sub-agent roles do the actual work with clear, non-overlapping responsibilities, so complex tasks progress in parallel instead of as one long linear thread. What is inside -… Read more →
Pros
- Clear four-role division of labor with a main-agent validation gate
- Prevents agent cascading by design, with per-role model selection and usage diagnostics
- Simple one-line install via the skills CLI
Cons
- Codex-only, no support for other coding agents
- Small project (145+ stars, 27 commits) with a thin track record
Pricing: Free · codex subagents agent-skill code-review multi-agent
21
Use ChatGPT, Cursor, OpenCode Zen and Antigravity models inside Claude Code
★ 185 · +6 this week · Apache-2.0 · updated 2026-09-29
6.5/10
What it does cc-multi-cli-plugin, also called Multi, lets one Claude Code session use models and coding harnesses from other providers. External models appear in the /model picker next to Claude's, and each provider can also run as a named worker subagent with progress, elapsed time and cancellation. Each provider keeps its own login, so you use the subscriptions you already… Read more →
Pros
- Each provider keeps its own login and native tools
- Separate claude-multi launcher leaves plain claude untouched
- Usage and receipts view across providers
Cons
- Only useful if you already pay for other providers
- Requires Node 24.12+ and several logins to set up
Pricing: Open source · claude-code-plugin multi-model subagents model-routing codex