GSA City Pair Program RFP 47QMCB25R0003
General Services Administration · 47QMCB25R0003. The government-wide airfare contract — 123 pages, with the fare tables kept page-cited rather than flattened. 443 modal-verb candidates, the deterministic tier. This is the pipeline’s output as delivered, unedited — page-cited rows, and the Coverage tab below stating what was and wasn’t parsed.
shredcheck—gsa-city-pair.xlsx
| ID | § | Pg | Requirement | Modal | Owner |
|---|---|---|---|---|---|
| R-005 | B.1 | 6 | 1. The contract carrier must furnish the services specified herein in accordance with the same services the carrier provides commercially to the general public in scheduled service… | must | |
| R-067 | C.6 | 30 | Connecting time must not exceed 210 minutes. | must | |
| R-080 | C.8 | 32 | The contract carrier must file contract fares under the “YCA”, “_CA”, “_ CB” or “_CP” designator. | must |
Three rows lifted verbatim from the workbook.
Download the workbook
Original solicitation: 47QMCB25R0003 (PDF, as fetched 2026-07-30)
Coverage — what this shred read, and what it didn’t
Extracted from the workbook’s own Coverage tab. Numbers below are the shred’s honest account of itself.
| Measure | Value |
|---|---|
| Pages parsed | 119 of 123 |
| Pages with no extractable text | 5, 48, 70, 73 |
| Modal candidates found | 443 |
| Added by full-document sweep | 82 |
| Dropped — sweep text not verbatim | 9 |
| Dropped — classified from context, not a candidate | 1 |
| QA corrections proposed but not applied | 0 |
| QA rows reviewed | 493 |
| QA corrections applied | 59 |
| QA review reached its output limit | no |
| Rows removed as carrying no obligation content | 0 |
| Classified binding | 445 |
| Excluded during classification | 33 |
| Referenced clauses detected | 137 |
| Attachments parsed inline | none detected |
| Attachments referenced but not parsed | ATTACHMENT 2, ATTACHMENT 1, ATTACHMENT 6, ATTACHMENT 4, ATTACHMENT 5, ATTACHMENT 3, ATTACHMENT 7, ATTACHMENT 8 |
| Standard-form pages not parsed | none detected |
| Rows depending on unread targets | R-001, R-024, R-058, R-061, R-069, R-072, R-272, R-358, R-360 |
| Rows whose page could not be uniquely confirmed | 18 |
| Rows whose text continues from the cited page | 0 |
| What this shred covered | Read 119 of 123 pages; 4 produced no extractable text and are not represented in this matrix (page 5, 48, 70, 73). 443 obligations were found by modal-verb matching and 82 more by a full-document pass, producing 493 matrix rows of which 445 are classified binding. 33 extracted passages were reviewed and produced no row because they carried no obligation. 9 passage(s) identified by the full-document pass were discarded because the text returned did not match the source exactly, and are not represented in this matrix. 1 obligation(s) were identified in surrounding context rather than in the passages being classified, and were not carried into the matrix. 18 row(s) quote text that occurs on more than one page, so the page shown for them could not be confirmed from that text alone; read those rows against their section heading rather than the page number. This document names 8 attachment(s) whose contents were not supplied: ATTACHMENT 2, ATTACHMENT 1, ATTACHMENT 6, ATTACHMENT 4, ATTACHMENT 5, ATTACHMENT 3, ATTACHMENT 7, ATTACHMENT 8. Obligations inside them are not in this matrix. 9 row(s) point at one of those unread targets and should be read as importing its obligations: R-001, R-024, R-058, R-061, R-069, R-072, R-272, R-358, R-360. Requirement text is reproduced as it appears in the source. Referenced clauses and standards are listed on the Referenced Clauses tab and are not expanded — reviewing what they import is expert work this shred does not perform. |
| Referenced clauses | Detected and listed, never expanded. A single citation can import further obligations; reviewing them is expert work this shred does not perform. |
| Confidence column | How many of three independent sweep runs found the row: High means all three, Medium two, Low one. Modal-verb rows read High by definition — that pass is deterministic and returned the same candidates on every run measured. In this workbook: 420 High, 12 Medium, 61 Low of 493 rows. The tiers say how a row was corroborated, not how likely it is to be right — no rate is quoted because the only grading on record was measured on a different distribution. |
| Build commit | f7729b4bcc04 |
| Generated (UTC) | 2026-09-04T16:15:19Z |
This sample also serves as one of the pipeline’s reference documents, so it shows the shred at its best — the fair test is your own document: drop any solicitation on the home page and the preview runs free in your browser. More samples.