convez.ai
MCP ServerDevOps & InfraestruturaAvançado
mickolasjae

Mick Applied AI Toolkit

mickolasjae/mick-applied-ai-toolkit

4 skills de produção: MCP audit, API resilience, evidence-gated CI e multi-model routing

Ver no GitHub
InicialPluginScore · de 0 a 100

Conjunto de 4 skills para Claude Code extraídos de produtos AI em produção: auditoria de MCP servers, resiliência em APIs Anthropic, CI com gates de evidência, e roteamento multi-modelo. Cada skill codifica uma disciplina testada com usuários pagantes.

Instalação

$npx convez-ai mickolasjae/mick-applied-ai-toolkit

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
57%
Último commit
há 5 meses
Versão
v0.1.1
Issues abertas
0
Contribuidores
2

Casos de uso

  • Auditar servidores MCP antes de submeter ao Anthropic Directory
  • Adicionar resiliência (round-robin keys, backoff, circuit breaker) a clientes Anthropic SDK
  • Implementar gates de CI com smoke tests, e2e, e verificação de artefatos antes de deploys
  • Rotear entre modelos Claude e GPT com seleção por tipo de tarefa e telemetria por chamada

Veredito

Conjunto maduro e opinionado de skills para engenheiros que rodam produtos AI em produção. As 4 disciplinas (auditoria, resiliência, CI, roteamento) resolvem problemas reais que quebram em escala; implementação é pragmática e testada em campo.

README

A tradução para o português deste README ainda está na fila.

mick-applied-ai-toolkit

Four Claude Code skills extracted from shipping production AI products end-to-end with the agentic coding loop. Each skill codifies a discipline that survives contact with paying users.

The skills

SkillWhat it doesSource pattern
mcp-directory-readinessAudits any MCP server against the Anthropic MCP Directory submission requirements; generates gap list + pre-filled form dataAnthropic's published submission flow + my own mcp-butterfly submission run
anthropic-api-resilienceDrops round-robin keys + 1s→5min exponential backoff + circuit breaker into any Anthropic SDK clientProduction reliability pattern surfaced in Anthropic's internal RL Studio codebase
evidence-gated-ciScaffolds smoketest + e2e + artifact-verified-mutation CI gate — no external mutation ships without a captured responseDiscipline I run across Tandem, Butterfly Security, LinguaMind, and mcp-butterfly
multi-model-routingCost-aware router with per-task-type model selection + per-call telemetryMulti-model routing pattern from Mercor's interview scoring codebase

Each skill is one folder under skills/ containing a SKILL.md with YAML frontmatter and supporting templates (TypeScript + Python drop-in code, reference docs, checklists).

Install (one skill)

Symlink any individual skill into your Claude Code skills directory:

ln -s "$(pwd)/skills/mcp-directory-readiness" ~/.claude/skills/mcp-directory-readiness

The skill auto-registers on the next Claude Code session start.

Install (all four)

for s in skills/*/; do
  ln -s "$(pwd)/$s" "$HOME/.claude/skills/$(basename $s)"
done

Use

Claude Code skills are auto-loaded once they live under ~/.claude/skills/. There is no /skill slash command — Claude invokes the matching skill itself (via the Skill tool) as soon as your prompt fires one of its triggers. So just describe the work in plain English:

  • "Audit my MCP server for the Anthropic Directory" → fires mcp-directory-readiness
  • "I'm getting 429s from the Claude API, add round-robin keys" → fires anthropic-api-resilience
  • "Add an evidence-gated CI pre-push gate to this project" → fires evidence-gated-ci
  • "Route between Claude and GPT for the cheapest model per task" → fires multi-model-routing

If you want to force a specific skill, name it explicitly in your prompt — e.g. "Use the multi-model-routing skill to scaffold a router for this app." The skill's name: field in SKILL.md is what Claude matches against.

See examples/ for a runnable scenario per skill (trigger prompt + expected agent behavior).

Why these four

I ship production AI products solo with Claude Code as the primary editor. These four skills are the disciplines that, when missing, broke something in front of real users.

  • mcp-directory-readiness — built while preparing mcp-butterfly for the Anthropic MCP Directory. The mechanical checks (tool annotations, OAuth posture, HTTPS) account for ~80% of submission rejections and are entirely automatable.
  • anthropic-api-resilience — built after watching LinguaMind silently drop user requests during a Claude API rate-limit window. Single-key clients fail badly at scale; round-robin + backoff converts it to invisible failover.
  • evidence-gated-ci — codifies what I do in every project's CLAUDE.md so it's portable. The load-bearing layer is artifact-verified mutations: no claim of external state change ships without a captured HTTP response on disk. Prevents the class of agent-fabricated state changes that pass review because nobody asked the external system whether the mutation actually happened.
  • multi-model-routing — built for the LinguaMind tutor (vocab extraction → Haiku, RAG synthesis → larger model). Routing by task-type with per-call cost telemetry compounds savings as DAU scales.

The portfolio behind the toolkit

  • Tandem — universal ebook + audiobook reader, live on iOS / iPadOS / macOS, visionOS in App Review. SwiftUI across four Apple platforms.
  • Butterfly Security — paying-customer SaaS for identity-infrastructure disaster recovery. Listed in the Okta Integration Network as an API Service Integration.
  • LinguaMind — AI language coach on Cloudflare Pages + D1 + Workers AI + Claude Haiku.
  • mcp-butterfly — remote streamable-HTTP MCP server on Cloudflare Workers + Durable Objects exposing Butterfly Security as Claude-callable tools.

Contributing

Issues and PRs welcome. Each skill has its own templates and reference files in the skills/<skill-name>/ folder — keep the structure consistent if you're adding a new skill.

License

MIT — see LICENSE.

Author

Mick Johnson · LinkedIn · GitHub

Ver o original no GitHub

Bom para

Relacionados