Prometheus Text Exposition Inspector

How to use Prometheus Text Exposition Inspector

Inspect local Prometheus text samples, escaped labels and selected histogram/summary consistency without scraping an endpoint. Review family metadata, duplicate series and cumulative bucket diagnostics in one local snapshot. Retain exact source tokens and separate finite binary64 samples from explicit NaN and infinity markers.

1. Provide prometheus text 0.0.4: legacy names, lf endings.
2. Choose the supported mode and run the tool. Review its assumptions and any reported errors.
3. Copy the result or download a separate output; reset to clear the current form.

Good to know

• 200 KB well-formed UTF-8, 5,000 LF-terminated lines, 8,192 characters per line, 2,000 samples and 32 labels per sample. Label values are at most 4,096 encoded characters; reports have at most 1,000 families/diagnostics and 500,000 output characters.
• Selected legacy ASCII metric/label names and decimal float profile only. UTF-8 quoted names, Go hexadecimal floats, alternate special-float spellings, OpenMetrics EOF/UNIT/exemplars, native histograms and protobuf are refused or unsupported. LF must terminate the file; CRLF is outside this profile. Only ASCII SP/TAB separate tokens; Unicode whitespace inside a quoted label or HELP text is retained.
• Decimal floats use binary64 and must remain finite without detected nonzero underflow. Explicit NaN, +Inf and -Inf remain separate source markers; timestamps are exact signed 64-bit integer milliseconds and remain strings.
• Histogram/summary checks group ordinary labels, inspect source ordering and check required count/sum metadata, +Inf equality, cumulative counts and quantile bounds. Count consistency is unavailable outside nonnegative safe binary64 integers; no rate, quantile re-estimation or counter-reset inference occurs.
• The inspector does not request a metrics endpoint, evaluate PromQL, ingest series or replace full promtool/collector conformance tests. A syntactically accepted or clean selected report is not proof of a healthy service.
• Inputs are processed on this device and are not automatically saved or uploaded.