Work · Evidence record
AI-led Vendor Catalog Normalization Across Messy SKUs: Four or More Checks Passed After One Correction
The completed WFT-059 synthetic field test reached 8/10 after one failure-only correction: 4 of five static checks passed for Catalog Normalization, while 1 check remained unresolved.
- Exact prompts and outputs
- One correction only
- Synthetic inputs disclosed
01 · The assignment
The task
normalize vendor catalog records across inconsistent SKU formats
02 · Scope before score
Test disclosures
Input disclosure
Synthetic blind-test input pack WFT-059: A procurement team will provide synthetic catalogs with aliases, unit mismatches, bundles, obsolete codes, and near-duplicate descriptions. Source facts: six fictional records WFT-059-C01 through WFT-059-C06; policy rules P1–P5; scores 17, 25, 58, 62, 74, and 91; a shared surname on C02/C05; and an incomplete identifier on WFT-059-C04. Governing rule card: all five written policy rules without adding an unstated tie-breaker. Apply the written rules in order; use descending disclosed scores only where ranking is requested; a shared name alone never proves identity; incomplete identifiers require abstention rather than invention. The pack deliberately contains no answer key, target ranking, expected classification, expected calculation result, or row-level decision. Every person, organization, record, statement, file, and identifier is fictional; no private, production, or live data was supplied.
Run disclosure
We ran one text-only synthetic benchmark WFT-059 for “normalize vendor catalog records across inconsistent SKU formats” in a Codex multi-agent session. We froze the first response, returned only its 2 detected check failures, accepted one corrected response, and scored both against the same five static checks. No external or live action occurred: nothing was sent, published, deployed, uploaded, submitted, purchased, booked, contacted, emailed, called, messaged, executed, or changed outside the fictional text fixtures. No external, live, or production action occurred. Per-case elapsed time was not instrumented during the batch session.
- Evidence mode
- Synthetic benchmark
- Run environment
- Codex multi-agent session
- Model disclosure
- Exact underlying model identifier not disclosed by the Codex session
03 · Verbatim input
Exact first prompt
The recorded session received the following prompt without silent additions.
Run bounded synthetic field test WFT-059. Task: normalize vendor catalog records across inconsistent SKU formats. Context: A procurement team will provide synthetic catalogs with aliases, unit mismatches, bundles, obsolete codes, and near-duplicate descriptions. Fictional source facts: six fictional records WFT-059-C01 through WFT-059-C06; policy rules P1–P5; scores 17, 25, 58, 62, 74, and 91; a shared surname on C02/C05; and an incomplete identifier on WFT-059-C04. Governing policy, formula, or rubric: all five written policy rules without adding an unstated tie-breaker. Apply the written rules in order; use descending disclosed scores only where ranking is requested; a shared name alone never proves identity; incomplete identifiers require abstention rather than invention. Produce a record-by-record decision matrix, ranked queue, and abstention log. Derive every row from the source facts and rule card; show calculations or criterion paths, cite supplied identifiers, state assumptions, surface any violated constraint, ambiguity, missing datum, or source conflict without presuming which row should pass, and abstain where evidence is incomplete. Do not infer an expected answer from the evidence plan and do not claim any live or external action. Planned verification after the frozen response: A canonical catalog and row-level mapping audit will verify identity matches, unit conversions, bundle handling, and preserved exceptions.04 · Baseline preserved
First result
The first response is retained before scoring or correction.
Frozen first response WFT-059 produced a record-by-record decision matrix, ranked queue, and abstention log for the task “normalize vendor catalog records across inconsistent SKU formats.” It treated the supplied pack as fictional and proposed this central handling: rank WFT-059-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-059-C04 until its identifier can be resolved. Concrete saved artifact row WFT-059-ROW1 reads: “WFT-059-C01 | rank WFT-059-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-059-C04 until its identifier can be resolved | source: supplied fictional pack | review state: first-draft.” The draft preserved the named records and separated supplied facts from assumptions; its inspectable work product also stated that no message, transaction, system change, or learner outcome had occurred. On the five predeclared static checks, it passed Catalog Normalization rule accuracy [WFT-059], Catalog Normalization exception handling [WFT-059], and Catalog Normalization source traceability [WFT-059]. The audit found concrete failures: for Catalog Normalization task fidelity [WFT-059], the saved draft did not connect WFT-059-C04 to the full boundary of “normalize vendor catalog records across inconsistent SKU formats”; for Catalog Normalization handoff usability [WFT-059], the saved draft left the record-by-record decision matrix, ranked queue, and abstention log without a complete reviewer handoff and acceptance marker. Those were evidence defects in the saved response, not inferred real-world failures, and they became the complete boundary of the single correction pass.Initial score: 6/10
05 · One pass only
Exact correction prompt
Only this single correction was allowed; there was no second repair pass.
Correct only these detected WFT-059 first-draft failures, using no new input or goal: 1) Catalog Normalization task fidelity [WFT-059] — the draft did not connect WFT-059-C04 to the full boundary of “normalize vendor catalog records across inconsistent SKU formats”; 2) Catalog Normalization handoff usability [WFT-059] — the draft left the record-by-record decision matrix, ranked queue, and abstention log without a complete reviewer handoff and acceptance marker.06 · Corrected output
Corrected final result
Corrected response WFT-059 retained the original fictional inputs, task boundary, and central decision: rank WFT-059-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-059-C04 until its identifier can be resolved. Concrete corrected artifact row WFT-059-ROW1 reads: “WFT-059-C01 | rank WFT-059-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-059-C04 until its identifier can be resolved | evidence locator: WFT-059-C01 | static status: 8/10 after one correction.” It changed only the detected failure areas, adding inspectable rows for Catalog Normalization handoff usability [WFT-059]. The frozen final text passed Catalog Normalization rule accuracy [WFT-059], Catalog Normalization exception handling [WFT-059], Catalog Normalization source traceability [WFT-059], and Catalog Normalization handoff usability [WFT-059] and still failed Catalog Normalization task fidelity [WFT-059]. The final record-by-record decision matrix, ranked queue, and abstention log therefore earned 8/10 from 4 passing checks; no second repair was attempted. This result reports only a bounded transcript-and-fixture evaluation. It does not claim that any workplace process, learner performance, account, device, service, or external system was actually changed, contacted, tested live, or improved.Final score: 8/10
07 · Five checks, two points each
Five-check record
The first and final statuses are textual as well as color coded. Each final pass is worth two points; the displayed verdict is tied to the final total.
| Check | First | Final | Evidence |
|---|---|---|---|
| Catalog Normalization task fidelity [WFT-059] | Fail | Fail | WFT-059 static check 1 inspected the saved wording for “Catalog Normalization task fidelity [WFT-059].” The first transcript did not contain enough support; the corrected transcript did not contain enough support. The comparison used only fictional fixture WFT-059-C04, the declared Catalog Normalization rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Catalog Normalization rule accuracy [WFT-059] | Pass | Pass | WFT-059 static check 2 inspected the saved wording for “Catalog Normalization rule accuracy [WFT-059].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-059-C04, the declared Catalog Normalization rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Catalog Normalization exception handling [WFT-059] | Pass | Pass | WFT-059 static check 3 inspected the saved wording for “Catalog Normalization exception handling [WFT-059].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-059-C04, the declared Catalog Normalization rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Catalog Normalization source traceability [WFT-059] | Pass | Pass | WFT-059 static check 4 inspected the saved wording for “Catalog Normalization source traceability [WFT-059].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-059-C04, the declared Catalog Normalization rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
| Catalog Normalization handoff usability [WFT-059] | Fail | Pass | WFT-059 static check 5 inspected the saved wording for “Catalog Normalization handoff usability [WFT-059].” The first transcript did not contain enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-059-C04, the declared Catalog Normalization rules, and the record’s frozen text; no live outcome or external action counted as evidence. |
08 · No cleanup by omission
What worked—and what failed
What worked
- WFT-059 kept “normalize vendor catalog records across inconsistent SKU formats” bounded to disclosed fictional inputs and preserved an auditable first-response snapshot.
- WFT-059 made the central handling—rank WFT-059-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-059-C04 until its identifier can be resolved—inspectable rather than implying unseen work.
- WFT-059 earned final passes for Catalog Normalization rule accuracy [WFT-059] and Catalog Normalization exception handling [WFT-059] under the same frozen scoring rules.
What failed or remained weak
- WFT-059 still lacked enough saved-text evidence for Catalog Normalization task fidelity [WFT-059]; the record leaves that final failure visible.
09 · Inspectable record
Evidence notes
A canonical catalog and row-level mapping audit will verify identity matches, unit conversions, bundle handling, and preserved exceptions.
- WFT-059 preserves the exact synthetic prompt, frozen first-response account, failure-only correction, corrected-response account, and five boolean decisions together.
- WFT-059 scores are arithmetic: 3 first-pass checks × 2 = 6/10; 4 final-pass checks × 2 = 8/10.
- WFT-059 evaluated only the text/static portion of the declared evidence plan—A canonical catalog and row-level mapping audit will verify identity matches, unit conversions, bundle handling, and preserved exceptions.—and did not fabricate a live artifact, external validator, or observed outcome.
10 · Boundary of the claim
Limitations
- WFT-059 is a synthetic benchmark, so its worked verdict measures fit to the disclosed fictional Catalog Normalization fixtures rather than effectiveness in a real workplace or learning setting.
- WFT-059 used one text-only Codex multi-agent session whose underlying model identifier was not disclosed; different prompts, models, fixtures, or human reviewers could produce different results.