Agents plugins

20 specialized agents for campaigns, content, and optimization

/scope locks Claude to the files you name; edits outside the scope are blocked.

/second-opinion has a different model critique Claude's last answer or plan.

/explore sends three subagents to investigate different parts of the codebase and merges their findings.

Fábrica de software autônoma com 71 agentes (40 core + 31 apoio) coordenados por Laura (Tech Lead). 19 skills cobrindo onboarding, especificação rastreável, mapa arquitetural, diagnóstico técnico de sistema existente com score e roadmap, threat model, grafo de conhecimento (Graph Engineering), validação contra SPEC fundamentada em arestas, execução conversacional/paralela, arco fechado de entrega com correção automática, revisão multi-agente, evals com rubrica como gate, handoff de design com verificação visual, lançamento com gates, otimização por ciclo de catraca, modernização por estrangulamento, auditoria (6 dimensões, incluindo Autonomia medida por telemetria) e evolução. PT-BR oficial. Roda em Claude Code, Codex CLI, OpenCode e Cursor. Para domínios regulados, use kairos-ai.

19 skills · 71 agents · hooks

Message your other local Claude Code, Codex, OpenCode and Antigravity sessions.

1 skill · 4 commands · 1 MCP server · hooks

17 skills. Agent and automation infrastructure: multi-agent protocols, tool curation, evals and runtimes.

17 skills

Read-only subagent that explains how code works, with file:line references.

Subagent that finds where code lives (superseded by the built-in Explore agent).

Subagent for structured code review with severity tiers.

Core agents and quality tools — planner, architect, reviewers, session management, verification, safety. Migrated from Everything Claude Code.

21 skills · 15 agents · hooks

Multi-agent team governance: self-bootstrap, role topology, boot receipts, delegation, CMUX delivery proof, heartbeat cadence, adversarial QA, finish audit. Auto-wires the @bradheitmann/odin-sentinel MCP server.

2 skills · 1 MCP server

Review and reorganize AI coding agent configuration and instruction files across Claude Code, Codex, and Copilot.

1 skill

Guide for creating new plugins in this repository with consistent structure and conventions.

1 skill

Send push notifications through ntfy when work finishes or needs a decision, on request or from another skill, including from remote and cloud sessions.

1 skill

80 specialized subagents - language experts, infra, review, security and domain specialists

80 agents

Mozart-conducted multi-agent orchestration: DELIVER, AUDIT, DIAGNOSE, INCIDENT, OPERATE, EVAL. Pluggable ticketing and documentation.

2 commands · 44 agents

Real-time, full-duplex messaging among Claude Code sessions and other coding agents. A session becomes a peer node on the SYM mesh: peers on one machine (loopback), a shared LAN (Bonjour), or across networks (WebSocket relay) discover each other and exchange typed cognitive signals — pushed as channel notifications where the host shows them, with sym_receive for the rest. Any MCP-capable agent joins the same mesh through the open @sym-bot/mesh-channel SDK; every peer keeps its own Ed25519 identity, proven on every session (MMP v2.0 Core Secure), its own SVAF admission, and local memory, on the open Mesh Memory Protocol (MMP).

1 MCP server

41-agent product development team for greenfield projects - includes managers and specialists across Architecture, Backend, Frontend, Marketing, Operations, Orchestrator, Product, and Quality domains

41 agents

Most agent infrastructure primarily adds capability to the execution plane. Maxey0 makes the contextual environment surrounding agentic execution an independently structured, routable, partitionable, and observable plane, while providing an engineering plane that can observe and tune both. The full stack: all three planes, ten slash commands, the role-agent slate, the Gate, and the Studio. Start here unless you know you want one plane.

17 skills · 10 commands · 5 agents · 3 MCP servers · hooks

Language, platform and concern specialists that routing dispatches to. Each verifies an API, a default, a limit or a deprecation against documentation before asserting it, returns what it consulted and what it could not confirm, and tries to break its own finding before reporting it.

35 agents

Shode House 4.1.0: 6 agent types, 20 skills, durable workflow and independent review.

20 skills · 6 commands · 6 agents

Build the thing: backend and frontend architects, AI and data engineers, DevOps, mobile, rapid prototyping, security engineering, and the small single-purpose agents for architecture, codegen, CI, docs and repo scanning.

25 agents

The Loop plane, alone. Route a task against 16 concepts, 83 skills, 67 agents and 84 hardened loops, and get the formation already built for it instead of an improvised maker/checker pair. Holds no window state at all, so it runs against a fixed context scheme with nothing else installed. 10 tools.

16 skills · 4 commands · 1 MCP server

Single-ticket implementation coordinator — fetches a Jira or Azure DevOps Boards ticket, routes to domain agents via multi-signal scoring, produces story-plan.md for approval, then sequentially invokes each domain agent with cross-domain context handoff, semantic memory retrieval, and a final cross-domain review

16 skills · 2 commands · 1 agent · hooks

Combat animation that is anatomically and martially right, for a game project that enables it in its own settings (never installed by install.sh). /move briefs a panel — anatomy, six martial areas with the declared tradition governing, animation craft, rig, Godot and three.js — and writes measurable constraints with their sources; /anim-review measures sampled clips against them and asks the owner's verdict; /anim-feedback keeps every verdict as an ADR that later runs read first; /reference fetches reference images with their licence; /meshy generates characters, weapons and props with Meshy, fetches them with their licence and judges them against the rig.

5 skills · 14 agents · hooks

Shared guardrails layer for all coding agent families — enforces safety, quality, governance, observability gates, and optional ChromaDB vector memory for semantic session recall

15 skills · 3 commands · hooks

Take it to market: content creation, growth hacking, app store optimisation, and per-channel specialists for Instagram, TikTok, Twitter/X, Reddit, WeChat, Xiaohongshu and Zhihu.

18 agents

Core frontend coding agent — config resolution, component-first routing, accessibility review, guidelines ingestion for React/Angular/Vue, and optional vector memory

13 skills · 2 commands · 1 agent · hooks

Run it once it is live: support response, analytics and executive reporting, finance tracking, infrastructure maintenance, legal and compliance checks, and data consolidation, extraction and distribution.

15 agents