On the comparison page, the "Uptime" and "Latency" columns have always beenLeave blank—Because nobody wants to take responsibility for made-up numbers. This tool fills those two columns: configure your endpoint list, then generate health-check configuration, health-check script, status badge, and uptime/latency table template. Numbers must be obtained via your own health checks, not copied.
One endpoint name + URL per line. MCP endpoints can also be probed—only HTTP status codes matter; no SSE parsing required.
—
—
—
—
1 · Using a JSON-RPC client to probe an MCP endpoint may incorrectly mark a live endpoint as dead.MCP’s streamable-http returns an SSE stream, not JSON. The correct approach:Judge solely by HTTP status code, both 200 and 402 are considered alive.
2 · Incorrect field names cause all entries to be permanently classified as "fluctuating."We set p.ok When a field actually exists but is written in the database as p.lastOk, and the result is all yellow. Field names must match exactly with the code that writes to the database.
3 · Status terms must be split into two columns: "action" and "neutral."Changing from "Online" to "Online" displays as "Restored Online," which is odd. We split it into CH_TOand CH_STATETwo sets (neutral term)
4 · Leave blank if no data is available; do not fabricate.The uptime/latency columns on the comparison page have always been empty—leaving them blank is more trustworthy than filling them with fabricated numbers until real health-check data is available.