Method 2026-10-06.temporal-evidence.1.
An observed timestamp difference does not establish clock error alone. Confirm the event pairing, reference clock, timestamp meaning and transport delays. The tool does not correct logs, synchronize clocks or infer a trusted source.
This pack contains the complete original parsed source values and must remain private. No source values or logs were changed. Reports summarize the first 100 rows and at most 400 characters per reference or finding. Complete inputs and findings remain in JSON, Excel and CSV.
JSON is a parsed-source backup, not an exact file-byte backup. Keep the unchanged original source separately, and restore it using Select file. Supply the frozen review inputs again. JSON and Excel preserve text; CSV is a spreadsheet-safe review export and can alter formula-like text for display.
Compare paired log and reference timestamps for the same known event. Calculate signed offsets in milliseconds and flag offsets above your selected tolerance.
Method revision 2026-10-06.temporal-evidence.1.
Compare integer milliseconds. Exact tolerance boundaries pass. Input timestamps require seconds, Z or a numeric offset within 14 hours, and at most three fractional digits. Years 1900 through 2200. Leap seconds, -00:00, named zones and surrounding whitespace are unsupported. This restricted profile is not a complete RFC 3339 validator.
References match exact source text. All duplicate references remain unresolved, including their first occurrence. Source rows count the header as row 1.
Original values remain unchanged. Complete parsed source and results are preserved in JSON.
An observed timestamp difference does not establish clock error alone. Confirm the event pairing, reference clock, timestamp meaning and transport delays. The tool does not correct logs, synchronize clocks or infer a trusted source.
