Dedicated tool guide
mcp_health
Check the authenticated MCP session, setup state, and advertised capabilities before starting a workflow.
Generated contract: apiVersion v1, schemaRevision 2. Examples use placeholders and do not call production.
mcp_health
read-onlyscope: noneConnection and setup probe: verifies the partner token and the processing chain end to end and reports under setup which prerequisites for a real send are met (sender profile, data processing agreement, balance, sandbox, approval, brand, letterhead, templates) and which step comes next. On a new account, setup.goldenPath lists the setup steps in order. Every step carries humanSummary (de, en), one sentence in everyday language for the person; tool names, nextTool, fixUrl and messageDe/messageEn are machine-facing. checkRender: true also checks whether rendering works right now (preview, thumbnail), for example before building a template that is then checked as an image. The answer arrives in about 3 seconds: a check over its time budget is listed under degraded and reports its cautious default.
Input contract
| Field | Type | Requirement | Meaning |
|---|---|---|---|
checkRender | boolean | optional | Also probes the render path (composer + rasterizer). Defaults to false. |
Contract-shaped example
mcp_health({ "checkRender": true })A failed or absent capability is a stop condition. The setup step keyed sandbox reports a per-account grant that is off by default; when it is unavailable, rehearse with dryRun: true instead. Never reinterpret a failed capability check as permission to send live mail.
For conditional fields and nested object limits, the generated complete tool reference remains authoritative. See errors and safe agent onboarding before implementing retries or a live send.
Trust and live status
Machine-readable service facts and the current operational status of the partner interface are publicly available at all times, no login needed.