Built for coding agents

Tudo que o dashboard faz existe em CLI não interativa, API e MCP. Este é o guia mínimo para Claude Code, Codex, Cursor e outros agentes. Versão machine-readable: /llms.txt

Deploy an existing application

# 1. Install the CLI
npm i -g simdeploy

# 2. Authenticate (token created at /dashboard/settings)
simdeploy login --token sd_live_...

# 3. Analyze (offline, no side effects)
simdeploy analyze --json

# 4. Non-interactive deploy — essential for agents
simdeploy deploy --yes

# 5. Verify (exit code != 0 on failure)
simdeploy status --json
simdeploy logs

Golden rules

Sempre --yes em automação; sem ele a CLI pede confirmação.
Sempre --json quando for interpretar a saída.
Builds run on the local machine; only the output goes to SimDeploy.
Estados de deployment: CREATED, ANALYZING, QUEUED, BUILDING, DEPLOYING, READY e as falhas tipadas ANALYSIS_FAILED, BUILD_FAILED, DEPLOY_FAILED.
Displayed costs are projections from price tables, never a real bill.

API v1

Auth: Authorization: Bearer sd_live_... — erros retornam {"error":{"code","message"}}.

GET /api/v1/me	validates token; returns organization and scopes
GET \| POST /api/v1/projects	list / create project {name}
GET \| DELETE /api/v1/projects/:id	detail / delete project
POST /api/v1/projects/:id/analyze	registers analysis; returns cost + planned route
GET \| POST /api/v1/projects/:id/deployments	list / deploy (multipart meta + artifact)
GET /api/v1/deployments/:id	status + pipeline events
GET /api/v1/deployments/:id/logs	structured logs (?format=text for plain text)
GET \| POST /api/v1/projects/:id/env	list keys / set variable
POST /api/v1/cost/estimate	standalone cost estimate

MCP

Servidor stdio simdeploy-mcp com as ferramentas analyze_project, estimate_cost, create_project, deploy_project, list_projects, get_project, get_deployment e get_logs. A autenticação reusa a da CLI.

{
  "mcpServers": {
    "simdeploy": { "command": "simdeploy-mcp" }
  }
}