convez.ai
AgenteDevOps & InfraestruturaIntermediário
LukeLalor

Claude Deploy

LukeLalor/claude-deploy

Deploy automatizado de Claude Agents via GitHub Actions

Ver no GitHub
InicialPluginScore · de 0 a 100

GitHub Action para deploy de Managed Agents do Claude direto do controle de versão. Implementa agentes como código, sincronizando configurações YAML com a plataforma Anthropic em cada push.

Instalação

$npx convez-ai LukeLalor/claude-deploy

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
5
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 5 meses
Versão
v1
Issues abertas
4
Contribuidores
1

Casos de uso

  • Versionamento de agentes gerenciados em repositório Git
  • CI/CD para atualizar agentes em produção ao modificar config
  • Sincronização de instruções e modelos entre ambientes

Veredito

Solução prática para quem precisa gerenciar Claude Agents em produção com disciplina de código. Integração limpa com GitHub Actions, mas requer familiaridade com workflows e configuração de secrets.

README

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

Deploy Claude Agent

A GitHub Action to deploy Claude Managed Agents from version control.

Why?

Managed agents are HTTP services. Like any other service, their configuration should live in version control. The fact that instructions are written in prose instead of code doesn't change this. Agents as code.

Example

Deploys your agent whenever its config file changes on main.

.github/workflows/deploy_agent.yml

name: Deploy Claude Agent

on:
  push:
    branches: [main]
    paths: [agents/**]

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Deploy agent
        id: deploy
        uses: LukeLalor/claude-deploy@v1
        with:
          config_file: agents/my-agent.yml
          anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}

      - run: echo "Deployed version ${{ steps.deploy.outputs.version }}"

agents/my-agent.yml

id: agent_abc123
name: My Agent
description: Does something useful
model:
  id: claude-opus-4-6
  speed: standard
system: You are a helpful assistant.

The agent to update is identified by the id field in the config file. Alternatively, omit id and pass the agent_id input instead (e.g. agent_id: ${{ vars.AGENT_ID }}). If both are set they must match.

To bootstrap your config file from an existing agent:

curl -s https://api.anthropic.com/v1/agents/$AGENT_ID -H 'anthropic-version: 2023-06-01' -H 'anthropic-beta: managed-agents-2026-04-01' -H "X-Api-Key: $ANTHROPIC_API_KEY" | yq -p json -o yaml 'del(.version)' > agents/my-agent.yml

Inputs

InputRequiredDescription
agent_idNoThe ID of the agent to update. Optional if the config file has an id
config_fileYesPath to the agent YAML config file (relative to repository root)
anthropic_api_keyYesAnthropic API key with permission to manage agents
allow_creationNoCreate a new agent when no agent ID is provided anywhere. Default false

Outputs

OutputDescription
agent_idThe ID of the agent created or updated
versionThe agent version after the update

Creating agents

With allow_creation: true, the action creates a new agent when no agent ID is provided via the agent_id input or the config file's id field. The new agent's ID is available in the agent_id output — add it to your config file (or store it as a repository variable) so subsequent runs update the same agent instead of creating another one.

Pinning a version

By default the action fetches the current version from the API automatically and passes it through with the update. To control versioning yourself, include version explicitly in your config file. Note that update operations will fail if the version is out of date, so opting in to this means you are responsible for keeping it in sync.

name: My Agent
model:
  id: claude-opus-4-6
  speed: standard
system: You are a helpful assistant.
version: 4
Ver o original no GitHub

Bom para

Relacionados