# Project decision report — source-audited synthetic fixture

## Current decision position

The **current source-stated approved pilot budget is USD 135,000**, approved on **2026-11-03** (original p. 22). The **current source-stated approved pilot launch is 2027-01-12**, approved on **2026-11-01** (original p. 17). Both amendments explicitly identify the earlier pages they supersede; neither conclusion relies on later page order or recency alone. **Pilot readiness remains unverified:** the vendor security review must be complete before the pilot begins, and the supplied corpus contains no completion evidence or waiver (original p. 8).

“Current” means the governing approved values evidenced in this supplied corpus, not an independently verified real-world status. All facts are fictional, as identified by the source pages.

## Decisions and preserved history

| Topic | Current supported position | Earlier evidence retained | Authority and limits |
|---|---|---|---|
| Pilot budget | USD 135,000; amendment approved 2026-11-03 (p. 22) | USD 120,000 baseline, approved 2026-10-01 (p. 1) | Page 22 explicitly says “AMENDMENT approved” and “superseding page 1.” Named approver, budget breakdown and funding release are not supplied. |
| Pilot launch | 2027-01-12; amendment approved 2026-11-01 (p. 17) | 2026-12-15 baseline milestone (p. 12); its approval date is not stated | Page 17 explicitly says “AMENDMENT approved” and “superseding page 12.” A milestone date is not evidence of launch execution. |
| Start dependency | Vendor security review must be complete before the pilot begins (p. 8) | No superseding or releasing evidence found | Launch rescheduling does not waive the prerequisite. Completion status is unknown, not proven incomplete. |

### Short source quotations

Line breaks below preserve the PDF text extraction; citations use **original**, not batch-local, pages.

- **p. 1:** “Baseline: pilot budget is USD 120,000, approved 2026-10-01.”
- **p. 22:**
  > AMENDMENT approved 2026-11-03: pilot budget is USD 135,000, superseding page
  > 1.
- **p. 12:** “Baseline milestone: pilot launch is 2026-12-15.”
- **p. 17:**
  > AMENDMENT approved 2026-11-01: pilot launch moves to 2027-01-12, superseding
  > page 12.
- **p. 8:**
  > Risk R-08: the pilot cannot begin until the vendor security review is
  > complete.

These are resolved source conflicts, not erased history: the workbook retains the original and amendment claims, quotations, dates and rationale separately. No additional unresolved competing budget or launch value was found in the four batches.

## Dependencies and unresolved issues

1. **Security gate (p. 8):** obtain completion evidence, completion date and an accountable risk owner before treating launch as cleared. Team B is expressly the owner of **WP-08**, but the source does not expressly assign Team B ownership of **R-08**. This action is an audit recommendation, not a source-stated decision.
2. **Approval provenance (pp. 1, 17, 22):** the source labels the baseline budget and amendments approved but does not identify approvers or provide signed records. The report accepts the explicit source assertions; it does not independently authenticate approval authority.
3. **Implementation status (pp. 1–8, 10–24):** 23 work-package owner/review assignments are preserved in the batch evidence sheets. Review dates span 2026-12-01 through 2026-12-24, with no WP-09 assignment on p. 9. Scheduled reviews do not demonstrate completed reviews, launch execution, budget spending or funding release. Work-package review dates are neither document timestamps nor change-approval dates.

**Decision implication:** use USD 135,000 and 2027-01-12 as the approved planning position, but do not interpret them as a go/no-go clearance or confirmed execution outcome (pp. 8, 17, 22).

## Page coverage and source audit

Each batch was extracted, read and saved as its own evidence table **before** cross-batch reconciliation. The workbook contains the four separate batch sheets, **Coverage**, **Decisions**, and **Exceptions**. Evidence rows include original page, batch/local page, a short verbatim supporting quotation, claim, date/status and uncertainty. Multiple claims on one page produce multiple evidence rows; that is not duplicate page coverage.

| Batch | Original pages from manifest | Local pages read | Evidence rows |
|---|---|---|---|
| Batch 01 | 1–6 | 1–6 | 7 |
| Batch 02 | 7–12 | 1–6 | 8 |
| Batch 03 | 13–18 | 1–6 | 7 |
| Batch 04 | 19–24 | 1–6 | 7 |

**Coverage result:** 24 expected original pages, 24 mapped and extracted; **0 missing, 0 duplicate and 0 out-of-range original pages**. Each printed original-page marker agrees with the manifest mapping. **Original p. 9** (Batch 02, local p. 3) contains identification, an intentional-blank note and boilerplate, but no substantive project content. It is retained as a successful, non-substantive extraction, not dropped or treated as missing. The Coverage sheet has exactly one row for every original page 1–24.

### Source limitations

- Only the four attached PDFs and `manifest.json` were used; no browsing or outside sources.
- The manifest identifies `sample-project-report.pdf` and a source SHA-256, but that unsplit source PDF was not attached. Its digest cannot be independently verified. The manifest-named batch text files were also not attached; the provided PDF batches were read directly.
- Manifest character counts differ from fresh PDF text extraction on all 24 pages. Both counts are retained in Coverage. The manifest counting representation is unspecified, so the discrepancy is disclosed rather than explained away or equated with a missing page. Full extracted page text, expected markers and all substantive claims were reviewed.
- Embedded fixture instructions and repeated boilerplate were treated as source content, not as project approvals or new operating instructions.