convez.ai
MCP ServerDevOps & InfraestruturaIntermediário
DopplerHQ

Doppler MCP Server

DopplerHQ/mcp-server

Integre gerenciamento de secrets Doppler com Claude via MCP

Ver no GitHub
InicialPluginScore · de 0 a 100

Servidor MCP que integra a API Doppler para gerenciamento de secrets e variáveis de ambiente com ferramentas geradas automaticamente a partir da especificação OpenAPI. Permite que assistentes de IA acessem e gerenciem segredos de forma segura com controle granular de permissões.

Instalação

$npx convez-ai DopplerHQ/mcp-server

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
7
Forks
3
Acompanhando
1
Saúde
25%
Último commit
há 1 mês
Versão
v1.0.5
Issues abertas
0
Contribuidores
3

Casos de uso

  • Automatizar gerenciamento de variáveis de ambiente em projetos
  • Consultar secrets e configurações de forma segura via IA
  • Integrar controle de secrets em workflows de DevOps
  • Auditar e revisar acesso a dados sensíveis com scoping automático

Veredito

MCP server sólido para integração Doppler em workflows Claude, com bom suporte a escopo de tokens e modo read-only. Experimental mas prático para equipes que já usam Doppler, exigindo atenção a práticas de segurança ao atribuir permissões.

README

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

Doppler MCP Server

MCP (Model Context Protocol) server that provides AI assistants with access to the Doppler API for secrets management.

[!IMPORTANT] The Doppler MCP Server is experimental and intended for development, testing, and evaluation purposes. Because outputs are non-deterministic and vary with the connected model, query, and server configuration, always use a token scoped only to the actions, projects, and environments you intend to allow, and review agentic output for alignment with your security and compliance requirements.

Quick Start

1. Authenticate:

npx @dopplerhq/mcp-server login

2. Add to your MCP client configuration (e.g., Claude Desktop):

{
  "mcpServers": {
    "doppler": {
      "command": "npx",
      "args": ["-y", "@dopplerhq/mcp-server"]
    }
  }
}

With Service Token

Alternatively, use a service token instead of logging in:

{
  "mcpServers": {
    "doppler": {
      "command": "npx",
      "args": ["-y", "@dopplerhq/mcp-server"],
      "env": {
        "DOPPLER_TOKEN": "<your-doppler-token>"
      }
    }
  }
}

See Service Tokens to create a config-scoped token.

Commands

login              Authenticate with Doppler (interactive)
logout             Clear cached auth credentials

Options

--read-only        Only expose read operations (GET endpoints)
--project <name>   Override auto-detected project
--config <name>    Override auto-detected config
--verbose, -v      Enable verbose logging to stderr
-h, --help         Show help message

Examples

Read-only mode

{
  "mcpServers": {
    "doppler": {
      "command": "npx",
      "args": ["-y", "@dopplerhq/mcp-server", "--read-only"]
    }
  }
}

Restrict to a specific project:

{
  "mcpServers": {
    "doppler": {
      "command": "npx",
      "args": ["-y", "@dopplerhq/mcp-server", "--project", "my-app"]
    }
  }
}

Restrict to a specific config:

{
  "mcpServers": {
    "doppler": {
      "command": "npx",
      "args": [
        "-y",
        "@dopplerhq/mcp-server",
        "--project",
        "my-app",
        "--config",
        "production"
      ]
    }
  }
}

Implicit Scope Detection

The server automatically detects scope based on your token's access:

Token AccessAuto-Detected Scope
Single project--project set automatically
Single project + single config--project and --config set automatically
Multiple projectsNo auto-detection (use CLI flags)

This means scoped service tokens work out of the box without any CLI flags:

{
  "mcpServers": {
    "doppler": {
      "command": "npx",
      "args": ["-y", "@dopplerhq/mcp-server", "--read-only"],
      "env": {
        "DOPPLER_TOKEN": "dp.st.xxx"
      }
    }
  }
}

CLI flags always take precedence over auto-detected values.

Security Best Practices

Use scoped service tokens, not CLI flags, for access control. The --project and --config flags provide a convenient UX layer but are not a substitute for proper token scoping. Always create service tokens with the minimum required permissions:

  1. Scope tokens to specific projects in the Doppler dashboard
  2. Use read-only tokens when write access isn't needed
  3. Combine read-only tokens with --read-only for defense in depth:
{
  "mcpServers": {
    "doppler": {
      "command": "npx",
      "args": ["-y", "@dopplerhq/mcp-server", "--read-only"],
      "env": {
        "DOPPLER_TOKEN": "dp.st.readonly_token"
      }
    }
  }
}

Note: The server cannot determine read/write permissions from the token, so if you're using a read-only token, add --read-only to only expose read tools. This prevents write tools from appearing in Claude's tool list and avoids failed API calls.

Available Tools

The server auto-generates tools from the Doppler OpenAPI spec. The tools available depend on your flags:

No flags (full access)

All Doppler API tools including:

  • Workplace: workplace_get, workplace_update
  • Users & Groups: users_list, groups_list, service_accounts_list
  • Projects: projects_list, projects_create, projects_get, projects_delete
  • Environments: environments_list, environments_create, environments_get
  • Configs: configs_list, configs_create, configs_get, configs_update, configs_lock
  • Secrets: secrets_list, secrets_get, secrets_update, secrets_download
  • Integrations: integrations_list, syncs_list, webhooks_list

With --read-only

Read-only tools only (GET operations). Write operations like _create, _update, _delete are not exposed.

With --project

Org-level tools are filtered out (workplace_*, activity_logs_*). The project parameter is auto-injected into tool calls that accept it.

With --config

Only config and secret management tools:

  • configs_get, configs_update, configs_lock, configs_unlock
  • secrets_list, secrets_get, secrets_update, secrets_download
  • config_logs_list, config_logs_rollback

Both project and config parameters are auto-injected.

Development

# Install dependencies
pnpm install

# Run tests
pnpm test

# Build
pnpm run build

# Run locally
DOPPLER_TOKEN=dp.xxx pnpm start

Branch and Release Flow

New work should branch from main and target main in PRs.

To release, push a tag in the format vX.X.X following semantic versioning. This triggers the publish workflow which builds and publishes to NPM.

Ver o original no GitHub

Bom para

Relacionados