OutageDeck 1.1.0: keyless dependency-status checks before debugging

[koko3tallah]

OutageDeck v1.1.0 is live in the Official MCP Registry as com.outagedeck/outagedeck-status.

The problem it targets is a small but expensive incident-response failure mode: an agent starts changing application code while AWS, Cloudflare, GitHub, OpenAI, or another dependency is already reporting an incident.

• Endpoint: https://outagedeck.com/api/mcp — remote Streamable HTTP
• Coverage: 172 cloud and SaaS vendors
• Public surface: 9 read-only tools, no API key or account
• Source model: each vendor official status feed, refreshed about every 10 minutes; OutageDeck does not infer outages from social reports or synthetic probes
• Server version: 1.1.0, protocol 2025-06-18

The public tools cover one-provider status, a whole-stack check for up to 12 providers, active incidents, full incident timelines, 7–90 day uptime history, provider search, outage reports, and MCP search / fetch documents with source links.

A typical agent prompt is:

Before debugging this deployment, check AWS, Cloudflare, GitHub, and OpenAI. If an official incident matches the component, symptom, region, and time window, show the incident timeline and avoid speculative code edits.

Claude Code setup:

claude mcp add --transport http outagedeck https://outagedeck.com/api/mcp

Generic client configuration:

{
  "mcpServers": {
    "outagedeck": {
      "url": "https://outagedeck.com/api/mcp"
    }
  }
}

Discovery and verification:

• Registry metadata repository: https://github.com/outagedeck/mcp
• Machine-readable manifest: https://outagedeck.com/.well-known/mcp.json
• Technical docs: https://outagedeck.com/developers/mcp?utm_source=github&utm_medium=discussion&utm_campaign=mcp_registry_showcase
• Independent connector scan: https://glama.ai/mcp/connectors/com.outagedeck/outagedeck-status

There are also five account-scoped tools for custom providers and alert management. They require an OutageDeck API key and are separate from the keyless public status tools. I would especially value feedback on this hybrid design: should public and account-scoped tools remain one discoverable server, or would client UX be clearer if they were split?

Disclosure: OutageDeck is maintained by the submitting organization. Public status checks require no signup or paid feature.