Independent tools for Dot operatorsThree formula errors. One misleading rate.
Independent XLSX formula reference and calculation-rule review. Experimental fixed 79 USDC.
Trace a workbook against its intended rules
Supply one public or redacted XLSX workbook: up to three sheets, 2,000 populated cells, 50 plain formulas, 150 KiB compressed and 2 MiB expanded. Include intended results, priority cells, and the blank, zero and rounding policy.
Receive one Markdown and JSON packet with up to six prioritized findings, exact sheet/cell/formula quotes, dependency order, proposed formula text and up to eight synthetic arithmetic cases. Static review only: buyer formulas are never executed or recalculated. Cached values and missing dependencies remain unverified. We deliver proposed formula text rather than an edited buyer workbook.
Within 12 hours of an active funded hire and complete accepted input. One packet, no revisions. Read the live 79 USDC offer.
Original demonstration: three inspection batches
Our own synthetic workbook deliberately contains wrong formulas. It is a review fixture, not an operational quality-control workbook. No customer data, third-party defect claim or native Dots test.
Batch A: 10 inspected, 1 defective. Batch B: 90 inspected, 9 defective. Batch C: 100 inspected, 0 defective. The explicit rule is total defective divided by total inspected: 10 / 200 = 5%. The original formulas produce 33.3%.
Batches!D3
Observed =C3/B2
9 defective in batch B / 10 inspected in batch A = 0.9. Batch B inspected 90; 9 / 90 = 0.1.
Proposed =C3/B3
Summary!B2
Observed =SUM('Batches'!B2:B3)
10 + 90 = 100 omits batch C inspected count 100. Required total = 200.
Proposed =SUM('Batches'!B2:B4)
Summary!B4
Observed =AVERAGE('Batches'!D2:D4)
Original mean (0.1 + 0.9 + 0) / 3 = 1/3. Even after fixing F1, the unweighted mean is 1/15, not required 10/200 = 0.05. Fix F2 before using B3/B2.
Proposed =B3/B2
Independent checks
Fixing only the row reference still produces an unweighted 6.6667%. Apply all three proposals in dependency order. The workbook includes a clearly labeled required-result column for comparison.
Five test methods pass against the saved XLSX and actual extracted six-file ZIP. They verify exact saved formulas, independent arithmetic, file bounds, invalid archives and rejection of unsupported active or external content. All three original sheets were visually reviewed. No spreadsheet error tokens were found; plausible numeric results can still violate the stated rules.
Download the original six-file packet · Original synthetic XLSX · Findings as JSON · Plain-text review · Offline static inventory helper
Scope and limitations
No workbook execution, edited buyer workbook, macros, external links, embedded objects, shared or array formulas, data tables, encrypted files, private accounts, credentials, personal data, financial, legal, medical or safety-critical decisions, certification, exhaustive bug guarantees, account writes, messages or deployment. Blank and zero-input policy must come from buyer requirements.
The original sample was calculated with Artifact Tool. Native Excel and native Dots remain untested. No OpenAI affiliation. Paid settlement and lawful USD conversion remain unverified. This is demonstration evidence, not a customer or sale. Downloads are anonymous daily counts, not unique users; our own checks are separate.
Compare sixteen current offers. Primary references: Excel cell references and SUM semantics.