benchme — in-pod Playwright vs Kitesurf (50 tasks)

Pre-registered rubric

• The "documentId" value in the final JSON object is the string "doc-101" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the key "documentId" whose value is a quoted JSON string, as the task requested.
• The response answers the request as asked: the value it reports is a document id (not only a title, URL or search snippet), and the response text presents it as the id of the returns and warranty policy document.
• The response text itself (not the run trace) names the document the id belongs to, e.g. its title as the vault shows it, so a reader can see which record the answer relies on.
• The response commits to its answer: it gives no alternative or conflicting "documentId" values, and if it mentions other candidate documents, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the vaultdocs app, the search), it does not also present a "documentId" value as found, and it never gives a placeholder such as "<id>" as the value.

Pre-registered rubric

• The "price" value in the final JSON object is the number 50.25, within ±0.005 (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "price" whose value is a bare JSON number (not a quoted string, and with no currency symbol or unit text inside the value).
• The response text presents the figure as the unit price of the Quenquen Renmar kit, the product the task names, and not as another product's price or a total for several units.
• The response text itself (not the run trace) identifies the product record the price comes from, e.g. by the product name and an identifier the warehouse app shows, so a reader can tell which record the figure relies on.
• The response commits to its answer: it gives no alternative or conflicting "price" values, and if it mentions other candidate products or prices, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the product), it does not also present a "price" value as found, and it never gives a placeholder such as "<number, USD>" as the value.

Pre-registered rubric

• The "customerCode" value in the final JSON object is the string "C-114" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the key "customerCode" whose value is a quoted JSON string, as the task requested.
• The response answers the request as asked: the value it reports is a customer code (not only a customer name or another identifier), and the response text presents that customer as gold-tier and based in Quillhaven, addressing each condition the task set.
• The response text itself (not the run trace) identifies the customer record the code came from, e.g. the customer's name as the warehouse app shows it, so a reader can see which record the answer relies on.
• The response commits to its answer: it gives no alternative or conflicting "customerCode" values, and if it mentions other candidate customers, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the customer), it does not also present a "customerCode" value as found, and it never gives a placeholder such as "<code>" as the value.

Pre-registered rubric

• The "qty" value in the final JSON object is the number 263 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "qty" whose value is a bare JSON number (not a quoted string, and with no unit text inside the value).
• The response text presents the figure as the current in-stock quantity of the Falnexmar Kael assembly (SKU ENC-1107) across all depots, as asked, and not as the count at an individual depot or a figure for another item.
• The response text itself (not the run trace) says which item record in the warehouse app the figure comes from, so a reader can trace where the number came from.
• The response commits to its answer: it gives no alternative or conflicting "qty" values, and if it mentions other candidate figures, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the item, covering the full scope asked), it does not also present a "qty" value as found, and it never gives a placeholder such as "<number>" as the value.

Pre-registered rubric

• The "qty" value in the final JSON object is the number 508 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "qty" whose value is a bare JSON number (not a quoted string, and with no unit text inside the value).
• The response text presents the figure as the current in-stock quantity of the Quenel Omren module (SKU TOO-1085) across all depots, as asked, and not as the count at an individual depot or a figure for another item.
• The response text itself (not the run trace) says which item record in the warehouse app the figure comes from, so a reader can trace where the number came from.
• The response commits to its answer: it gives no alternative or conflicting "qty" values, and if it mentions other candidate figures, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the item, covering the full scope asked), it does not also present a "qty" value as found, and it never gives a placeholder such as "<number>" as the value.

Pre-registered rubric

• The "qty" value in the final JSON object is the number 912 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "qty" whose value is a bare JSON number (not a quoted string, and with no unit text inside the value).
• The response text presents the figure as the current in-stock quantity of the Zyom Renom kit (SKU FAS-1000) across all depots, as asked, and not as the count at an individual depot or a figure for another item.
• The response text itself (not the run trace) says which item record in the warehouse app the figure comes from, so a reader can trace where the number came from.
• The response commits to its answer: it gives no alternative or conflicting "qty" values, and if it mentions other candidate figures, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the item, covering the full scope asked), it does not also present a "qty" value as found, and it never gives a placeholder such as "<number>" as the value.

Pre-registered rubric

• The "qty" value in the final JSON object is the number 371 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "qty" whose value is a bare JSON number (not a quoted string, and with no unit text inside the value).
• The response text presents the figure as the quantity of the Loom Loquen kit (SKU OPT-1097) held at depot OST (Zyvi depot 2), as asked, and not as a total across depots or a figure for another depot or item.
• The response text itself (not the run trace) identifies the item and the depot entry the figure was read from, so a reader can trace where the number came from.
• The response commits to its answer: it gives no alternative or conflicting "qty" values, and if it mentions other candidate figures, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the item at that depot), it does not also present a "qty" value as found, and it never gives a placeholder such as "<number>" as the value.

Pre-registered rubric

• The "qty" value in the final JSON object is the number 89 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "qty" whose value is a bare JSON number (not a quoted string, and with no unit text inside the value).
• The response text presents the figure as the quantity of the Loloom Falvi unit (SKU FAS-1040) held at depot QHV (Thoomsil depot 3), as asked, and not as a total across depots or a figure for another depot or item.
• The response text itself (not the run trace) identifies the item and the depot entry the figure was read from, so a reader can trace where the number came from.
• The response commits to its answer: it gives no alternative or conflicting "qty" values, and if it mentions other candidate figures, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the item at that depot), it does not also present a "qty" value as found, and it never gives a placeholder such as "<number>" as the value.

Pre-registered rubric

• The "price" value in the final JSON object is the number 23.5, within ±0.005 (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "price" whose value is a bare JSON number (not a quoted string, and with no currency symbol or unit text inside the value).
• The response text presents the figure as the unit price of the Lolo Vivi pack, the product the task names, and not as another product's price or a total for several units.
• The response text itself (not the run trace) identifies the product record the price comes from, e.g. by the product name and an identifier the warehouse app shows, so a reader can tell which record the figure relies on.
• The response commits to its answer: it gives no alternative or conflicting "price" values, and if it mentions other candidate products or prices, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the product), it does not also present a "price" value as found, and it never gives a placeholder such as "<number, USD>" as the value.

Pre-registered rubric

• The "price" value in the final JSON object is the number 148.25, within ±0.005 (a JSON number, not a quoted string).
• The final reply contains a JSON object with the key "price" whose value is a bare JSON number (not a quoted string, and with no currency symbol or unit text inside the value).
• The response text presents the figure as the unit price of the Vizynex Renquen pack, the product the task names, and not as another product's price or a total for several units.
• The response text itself (not the run trace) identifies the product record the price comes from, e.g. by the product name and an identifier the warehouse app shows, so a reader can tell which record the figure relies on.
• The response commits to its answer: it gives no alternative or conflicting "price" values, and if it mentions other candidate products or prices, it says why it did not use them.
• The response is candid about gaps: if it says a step failed or was incomplete (creating the workspace, opening the warehouse app, finding the product), it does not also present a "price" value as found, and it never gives a placeholder such as "<number, USD>" as the value.

Pre-registered rubric

• The "customerCode" value in the final JSON object is the string "C-115" (case-insensitive, surrounding whitespace ignored).
• The final reply includes a JSON object of the shape the task requested: the key "customerCode" with its value, and no other keys in that object.
• The "customerCode" value is a customer code string — not a customer name, a sentence, a list of codes or the placeholder "<code>".
• The response shows its basis: it names the customer the code belongs to and states that this customer is platinum-tier and based in Karrow, addressing the conditions the task set.
• Any value for "customerCode" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "tier" value in the final JSON object is the string "gold" (case-insensitive, surrounding whitespace ignored).
• The final reply includes a JSON object of the shape the task requested: the key "tier" with its value, and no other keys in that object.
• The "tier" value is a tier label string on its own — not a sentence, a list of tiers or the placeholder "<tier>".
• The response names the customer the task asked about (Quenfalsil Industrial) together with the tier, so the value is clearly tied to that customer and not another.
• Any value for "tier" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "tier" value in the final JSON object is the string "gold" (case-insensitive, surrounding whitespace ignored).
• The final reply includes a JSON object of the shape the task requested: the key "tier" with its value, and no other keys in that object.
• The "tier" value is a tier label string on its own — not a sentence, a list of tiers or the placeholder "<tier>".
• The response names the customer the task asked about (Kasilvi Systems) together with the tier, so the value is clearly tied to that customer and not another.
• Any value for "tier" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "count" value in the final JSON object is the number 4 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object of the shape the task requested: the key "count" with its value, and no other keys in that object.
• The "count" value is a bare JSON number (a whole number) — not a quoted string, a range, a word or an expression.
• The response makes clear the count is of the open orders of Vitru Industrial (C-127) and shows its basis: it identifies the orders it counted (e.g. by order number) or says how it decided which of that customer's orders were open.
• Any value for "count" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "count" value in the final JSON object is the number 2 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object of the shape the task requested: the key "count" with its value, and no other keys in that object.
• The "count" value is a bare JSON number (a whole number) — not a quoted string, a range, a word or an expression.
• The response makes clear the count is of the open orders of Thomarom Systems (C-115) and shows its basis: it identifies the orders it counted (e.g. by order number) or says how it decided which of that customer's orders were open.
• Any value for "count" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "status" value in the final JSON object is the string "shipped" (case-insensitive, surrounding whitespace ignored).
• The final reply includes a JSON object of the shape the task requested: the key "status" with its value, and no other keys in that object.
• The "status" value is a status label string on its own — not a sentence, a list of statuses or the placeholder "<status>".
• The response names the order it looked up (SO-20003) alongside the status, so the value is clearly tied to that order and not another.
• Any value for "status" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "status" value in the final JSON object is the string "shipped" (case-insensitive, surrounding whitespace ignored).
• The final reply includes a JSON object of the shape the task requested: the key "status" with its value, and no other keys in that object.
• The "status" value is a status label string on its own — not a sentence, a list of statuses or the placeholder "<status>".
• The response names the order it looked up (SO-20040) alongside the status, so the value is clearly tied to that order and not another.
• Any value for "status" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "status" value in the final JSON object is the string "cancelled" (case-insensitive, surrounding whitespace ignored).
• The final reply includes a JSON object of the shape the task requested: the key "status" with its value, and no other keys in that object.
• The "status" value is a status label string on its own — not a sentence, a list of statuses or the placeholder "<status>".
• The response names the order it looked up (SO-20167) alongside the status, so the value is clearly tied to that order and not another.
• Any value for "status" stated in the prose matches the value in the JSON object; the response gives no other, conflicting value.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "total" value in the final JSON object is the number 7093.5, within ±0.005 (a JSON number, not a quoted string).
• The final reply includes a JSON object of the shape the task requested: the key "total" with its value, and no other keys in that object.
• The "total" value is a bare JSON number — not a quoted string, and with no "$", currency code, unit or thousands separator inside the JSON.
• The response shows the basis for the total: it names the order it read (SO-20296) and either lists the line figures it summed or says which of that order's own figures it relied on.
• Any total stated in the prose matches the JSON "total" (a "$" sign or thousands separators in the prose are fine); the response gives no other, conflicting total.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "total" value in the final JSON object is the number 8691.75, within ±0.005 (a JSON number, not a quoted string).
• The final reply includes a JSON object of the shape the task requested: the key "total" with its value, and no other keys in that object.
• The "total" value is a bare JSON number — not a quoted string, and with no "$", currency code, unit or thousands separator inside the JSON.
• The response shows the basis for the total: it names the order it read (SO-20102) and either lists the line figures it summed or says which of that order's own figures it relied on.
• Any total stated in the prose matches the JSON "total" (a "$" sign or thousands separators in the prose are fine); the response gives no other, conflicting total.
• The response does not present a guessed, assumed or placeholder value as if it were read from the warehouse app: if a step failed or the record it needed could not be read, it says so explicitly and does not present the value as confirmed.

Pre-registered rubric

• The "total" value in the final JSON object is the number 16555.25, within ±0.005 (a JSON number, not a quoted string).
• The final reply contains a JSON object with the field "total" whose value is a single JSON number (not a quoted string, a range, or text with a currency symbol or thousands separators).
• The response identifies its value as the total of order SO-20156 across all of that order's lines (it names that order), not as a single line's amount or another order's figure.
• The response commits to a single value for "total"; it does not offer alternative, hedged or conflicting values.
• If the response shows supporting figures (line quantities, unit prices, line amounts or subtotals), they are arithmetically consistent with the "total" it reports.
• If the response reports that it could not read every line of order SO-20156, it marks its total as partial or unverified rather than presenting it as complete.

Pre-registered rubric

• The "qty" value in the final JSON object is the number 9 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the field "qty" whose value is a single JSON number (not a quoted string, a range, or text such as "units").
• The response identifies its value as the quantity of SKU ENC-1075 on order SO-20060 (it names that SKU or order), not as a different figure such as the order's total units or stock on hand.
• The response commits to a single value for "qty"; it does not offer alternative, hedged or conflicting values.
• If the response reports any difficulty finding SKU ENC-1075 on order SO-20060 or reading that line, it marks its quantity as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "qty" value in the final JSON object is the number 12 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the field "qty" whose value is a single JSON number (not a quoted string, a range, or text such as "units").
• The response identifies its value as the quantity of SKU CAB-1034 on order SO-20118 (it names that SKU or order), not as a different figure such as the order's total units or stock on hand.
• The response commits to a single value for "qty"; it does not offer alternative, hedged or conflicting values.
• If the response reports any difficulty finding SKU CAB-1034 on order SO-20118 or reading that line, it marks its quantity as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "toCode" value in the final JSON object is the string "VLM" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "toCode" whose value is a depot code given as a JSON string (not a depot's full name, its city, or a sentence).
• The response identifies the code as the destination of transfer TR-532 (it names that transfer), not as that transfer's origin or another transfer's destination.
• The response commits to a single depot code for "toCode"; it does not offer alternative, hedged or conflicting depot codes.
• If the response reports any difficulty finding transfer TR-532 or reading its destination, it marks its depot code as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "location" value in the final JSON object is the string "QHV" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "location" whose value is a depot code given as a JSON string (not a depot's full name, its city, or a sentence).
• The response frames its answer as the depot with the most units of SKU FLU-1055 specifically (it names that SKU or product), not as the depot with the most stock overall.
• The response commits to a single depot code for "location"; it does not offer alternative, hedged or conflicting depot codes.
• The response states the basis for its choice: the stock figures for SKU FLU-1055 it compared across depots, rather than a bare depot code with no stated basis.
• If the response reports a tie or incomplete stock data for SKU FLU-1055, it marks its pick as uncertain rather than presenting it as definitive.

Pre-registered rubric

• The "sku" value in the final JSON object is the string "ENC-1043" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "sku" whose value is a single SKU code given as a JSON string (not a product name, a price, or a list).
• The response frames its pick as the enclosures-category product with the highest unit price (it states unit price as the ranking criterion), not as a pick by stock level, order volume or popularity.
• The response commits to a single SKU code for "sku"; it does not offer alternative, hedged or conflicting SKU codes.
• The response states the basis for its pick: the unit price it found for the chosen product and the prices it compared it against, rather than a bare SKU with no stated basis.
• If the response reports a price tie or that it could not compare every enclosures product, it marks its pick as uncertain rather than presenting it as definitive.

Pre-registered rubric

• The "city" value in the final JSON object is the string "Tessalind" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "city" whose value is a city name given as a JSON string (not a depot code, a full street address, or a sentence).
• The response identifies the city as the location of depot QHV (it names that depot by code or name), not as the city of a different depot.
• The response commits to a single city for "city"; it does not offer alternative, hedged or conflicting cities.
• If the response reports any difficulty finding depot QHV, reading its city, or conflicting location details, it marks its city as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "requester" value in the final JSON object is the string "C-124" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "requester" whose value is a customer code given as a JSON string (not a customer's name, an email address, a support agent, or a sentence).
• The response identifies the customer as the reporter of ticket HD-5065 (it names that ticket), not as the ticket's assignee or the customer on a different ticket.
• The response commits to a single customer code for "requester"; it does not offer alternative, hedged or conflicting customer codes.
• If the response reports any difficulty finding ticket HD-5065 or reading its requester, it marks its customer code as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "requester" value in the final JSON object is the string "C-114" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "requester" whose value is a customer code given as a JSON string (not a customer's name, an email address, a support agent, or a sentence).
• The response identifies the customer as the reporter of ticket HD-5085 (it names that ticket), not as the ticket's assignee or the customer on a different ticket.
• The response commits to a single customer code for "requester"; it does not offer alternative, hedged or conflicting customer codes.
• If the response reports any difficulty finding ticket HD-5085 or reading its requester, it marks its customer code as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "requester" value in the final JSON object is the string "C-132" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "requester" whose value is a customer code given as a JSON string (not a customer's name, an email address, a support agent, or a sentence).
• The response identifies the customer as the reporter of ticket HD-5115 (it names that ticket), not as the ticket's assignee or the customer on a different ticket.
• The response commits to a single customer code for "requester"; it does not offer alternative, hedged or conflicting customer codes.
• If the response reports any difficulty finding ticket HD-5115 or reading its requester, it marks its customer code as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "priority" value in the final JSON object is the string "high" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "priority" whose value is a priority label given as a JSON string (not a list, a response or resolution time, or a sentence).
• The response presents the value as the priority of ticket HD-5128 itself (it names that ticket), not the ticket's status or the priority of a different ticket.
• The response commits to its value for "priority": the JSON and any prose state the same priority, with no alternative, hedged or conflicting priorities.
• If the response reports any difficulty finding ticket HD-5128 or reading its priority, it marks its priority as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "priority" value in the final JSON object is the string "low" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "priority" whose value is a priority label given as a JSON string (not a list, a response or resolution time, or a sentence).
• The response presents the value as the priority of ticket HD-5136 itself (it names that ticket), not the ticket's status or the priority of a different ticket.
• The response commits to its value for "priority": the JSON and any prose state the same priority, with no alternative, hedged or conflicting priorities.
• If the response reports any difficulty finding ticket HD-5136 or reading its priority, it marks its priority as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "status" value in the final JSON object is the string "open" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "status" whose value is a ticket status label given as a JSON string (not a date, a list, or a sentence).
• The response presents the value as the status of ticket HD-5129 itself (it names that ticket), not the ticket's priority, the status of another record the ticket mentions, or a different ticket's status.
• The response commits to its value for "status": the JSON and any prose state the same status, with no alternative, hedged or conflicting statuses.
• If the response reports any difficulty finding ticket HD-5129 or reading its status, it marks its status as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "status" value in the final JSON object is the string "open" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "status" whose value is a ticket status label given as a JSON string (not a date, a list, or a sentence).
• The response presents the value as the status of ticket HD-5091 itself (it names that ticket), not the ticket's priority, the status of another record the ticket mentions, or a different ticket's status.
• The response commits to its value for "status": the JSON and any prose state the same status, with no alternative, hedged or conflicting statuses.
• If the response reports any difficulty finding ticket HD-5091 or reading its status, it marks its status as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "assigneeCode" value in the final JSON object is the string "AG-14" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "assigneeCode" whose value is an agent code given as a JSON string, as the task asked (not the agent's name, a team, or a sentence).
• The response presents the agent as the assignee of ticket HD-5025 (it names that ticket), not the ticket's requester (the customer) or the agent on a different ticket.
• The response commits to its value for "assigneeCode": the JSON and any prose state the same agent code, with no alternative, hedged or conflicting agent codes.
• If the response reports that ticket HD-5025 shows no assignee, or any difficulty finding it or reading its assignee, it does not present an agent code for it as confirmed.

Pre-registered rubric

• The "assigneeCode" value in the final JSON object is the string "AG-16" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "assigneeCode" whose value is an agent code given as a JSON string, as the task asked (not the agent's name, a team, or a sentence).
• The response presents the agent as the assignee of ticket HD-5069 (it names that ticket), not the ticket's requester (the customer) or the agent on a different ticket.
• The response commits to its value for "assigneeCode": the JSON and any prose state the same agent code, with no alternative, hedged or conflicting agent codes.
• If the response reports that ticket HD-5069 shows no assignee, or any difficulty finding it or reading its assignee, it does not present an agent code for it as confirmed.

Pre-registered rubric

• The "assigneeCode" value in the final JSON object is the string "AG-15" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "assigneeCode" whose value is an agent code given as a JSON string, as the task asked (not the agent's name, a team, or a sentence).
• The response presents the agent as the assignee of ticket HD-5063 (it names that ticket), not the ticket's requester (the customer) or the agent on a different ticket.
• The response commits to its value for "assigneeCode": the JSON and any prose state the same agent code, with no alternative, hedged or conflicting agent codes.
• If the response reports that ticket HD-5063 shows no assignee, or any difficulty finding it or reading its assignee, it does not present an agent code for it as confirmed.

Pre-registered rubric

• The "team" value in the final JSON object is the string "tier-2" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "team" whose value is a team name given as a JSON string (not an agent code, a person's name, a list, or a sentence).
• The response presents the value as the team of support agent Lior Kowalczyk (it names that agent), not a different agent's team.
• The response commits to its value for "team": the JSON and any prose state the same team, with no alternative, hedged or conflicting teams.
• If the response reports any difficulty finding support agent Lior Kowalczyk or reading that agent's team, it marks its team as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "team" value in the final JSON object is the string "billing" (case-insensitive, surrounding whitespace ignored).
• The final reply contains a JSON object with the field "team" whose value is a team name given as a JSON string (not an agent code, a person's name, a list, or a sentence).
• The response presents the value as the team of support agent Quill Ferrante (it names that agent), not a different agent's team.
• The response commits to its value for "team": the JSON and any prose state the same team, with no alternative, hedged or conflicting teams.
• If the response reports any difficulty finding support agent Quill Ferrante or reading that agent's team, it marks its team as unverified rather than presenting it as confirmed.

Pre-registered rubric

• The "count" value in the final JSON object is the number 2 exactly (a JSON number, not a quoted string).
• The final reply contains a JSON object with the field "count" whose value is a bare JSON number (not a quoted string, a range, or a list of tickets in place of the number).
• The response presents the number as the count of helpdesk tickets reported by customer Omthonex Supply (C-131) (it names that customer), not tickets handled by an agent or another customer's tickets; if it counted only some of that customer's tickets, it says so.
• The response commits to its value for "count": the JSON and any prose state the same count, and if it lists the tickets it counted, the number listed equals that count.
• The response states the basis for its count: it identifies the tickets it counted (e.g. by ticket number) or the record or listing it took the figure from, rather than giving a bare number.
• If the response reports that it may not have seen all of customer C-131's tickets, it marks its count as partial or unverified rather than presenting it as complete.

Pre-registered rubric

• The "documentId" value in the final JSON object is the string "doc-130" (case-insensitive, surrounding whitespace ignored).
• The final reply includes a JSON object with the field "documentId" whose value is a single JSON string holding a document id — not a document title, a URL, a search snippet, a list, or the placeholder "<id>".
• The response ties the id to the document the task asked about: alongside the id it names that document (e.g. its title as shown in the vault), and the document it names is operations memo 2: rentruzy retrospective, not a different memo.
• The response commits to a single id: any document id it gives as its answer in the prose matches the JSON value, and it does not leave a choice between several candidate ids to the reader.
• The response does not present a guessed, assumed or placeholder id as if it were read from the vault: if it reports that a step failed or that the document could not be found or read, it does not present the id as confirmed.

Pre-registered rubric

• The "days" value in the final JSON object is the number 2 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "days" whose value is a bare JSON number — not a quoted string, a range, a word, or text with units inside the value.
• The response labels the value it reports as the standard supplier lead time for the Zyom Renom kit (SKU FAS-1000) — it names that product or SKU — not as a different kind of figure or another product's figure.
• The response shows its basis: it names the vault document it read the figure from (e.g. its title as shown in the vault), so a reader can see that the answer relies on the datasheet for SKU FAS-1000.
• The response commits to a single value for "days": any value it gives as its answer in the prose matches the JSON value, and it offers no alternative, hedged or conflicting values.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the datasheet for SKU FAS-1000 could not be found or read, it does not present the value as confirmed.

Pre-registered rubric

• The "days" value in the final JSON object is the number 6 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "days" whose value is a bare JSON number — not a quoted string, a range, a word, or text with units inside the value.
• The response labels the value it reports as the standard supplier lead time for the Lofalel Zynex assembly (SKU ADH-1054) — it names that product or SKU — not as a different kind of figure or another product's figure.
• The response shows its basis: it names the vault document it read the figure from (e.g. its title as shown in the vault), so a reader can see that the answer relies on the datasheet for SKU ADH-1054.
• The response commits to a single value for "days": any value it gives as its answer in the prose matches the JSON value, and it offers no alternative, hedged or conflicting values.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the datasheet for SKU ADH-1054 could not be found or read, it does not present the value as confirmed.

Pre-registered rubric

• The "moq" value in the final JSON object is the number 5 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "moq" whose value is a bare JSON number — not a quoted string, a range, a word, or text with units inside the value.
• The response labels the value it reports as the minimum order quantity for the Baelfal Marren pack (SKU CAB-1018) — it names that product or SKU — not as a different kind of figure or another product's figure.
• The response shows its basis: it names the vault document it read the figure from (e.g. its title as shown in the vault), so a reader can see that the answer relies on the datasheet for SKU CAB-1018.
• The response commits to a single value for "moq": any value it gives as its answer in the prose matches the JSON value, and it offers no alternative, hedged or conflicting values.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the datasheet for SKU CAB-1018 could not be found or read, it does not present the value as confirmed.

Pre-registered rubric

• The "percent" value in the final JSON object is the number 9 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "percent" whose value is a bare JSON number expressing a percentage (not a decimal fraction) — not a quoted string, a range, or text with a % sign or units inside the value.
• The response labels the value it reports as the discount off list price for platinum customers under the customer tier discounts policy (it names the platinum tier), not as another tier's discount or a different term of that policy.
• The response shows its basis: it names the vault document it read the figure from (e.g. its title as shown in the vault), so a reader can see that the answer relies on the customer tier discounts policy.
• The response commits to a single value for "percent": any value it gives as its answer in the prose matches the JSON value, and it offers no alternative, hedged or conflicting values.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the customer tier discounts policy could not be found or read, it does not present the value as confirmed.

Pre-registered rubric

• The "days" value in the final JSON object is the number 14 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "days" whose value is a bare JSON number — not a quoted string, a range, a word, or text with units inside the value.
• The response labels the value it reports as the number of days after delivery within which customers may return unused goods, not as a time limit for something else or a figure in other units.
• The response shows its basis: it names the vault document it read the figure from (e.g. its title as shown in the vault), so a reader can see that the answer relies on the returns and warranty policy.
• The response commits to a single value for "days": any value it gives as its answer in the prose matches the JSON value, and it offers no alternative, hedged or conflicting values.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the returns and warranty policy could not be found or read, it does not present the value as confirmed.

Pre-registered rubric

• The "hour" value in the final JSON object is the number 19 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "hour" whose value is a bare JSON number giving the hour as a whole number — not a quoted string, a time with minutes, an am/pm time, or a range.
• The response labels the hour it reports as the time until which Thoomsil depot 3 (QHV) dispatches goods — it names that depot by name or code — not as some other hour of that depot or another depot's hour.
• The response shows its basis: it names the vault document it read the figure from (e.g. its title as shown in the vault), so a reader can see that the answer relies on the depot operating hours policy.
• Any hour the response gives as its answer in the prose matches the JSON value and is expressed on a 24-hour clock; the response offers no alternative, hedged or conflicting hours.
• The response does not present a guessed, assumed or placeholder hour as if it were read from the vault: if it reports that a step failed or that the depot operating hours policy could not be found or read, it does not present the hour as confirmed.

Pre-registered rubric

• The "months" value in the final JSON object is the number 3 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "months" whose value is a bare JSON number — not a quoted string, a range, a word, or text with units inside the value.
• The response labels the value it reports as the number of months until Trumar Systems' next review — it names that account — not as a calendar date, a duration in other units, or another account's review timing.
• The response shows its basis: it names the vault document it read the figure from (e.g. its title as shown in the vault), so a reader can see that the answer relies on the account review notes for Trumar Systems.
• The response commits to a single value for "months": any value it gives as its answer in the prose matches the JSON value, and it offers no alternative, hedged or conflicting values.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the account review notes for Trumar Systems could not be found or read, it does not present the value as confirmed.

Pre-registered rubric

• The "employees" value in the final JSON object is the number 614 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "employees" whose value is a bare JSON number — not a quoted string, a range, a word, or text with units inside the value.
• The response labels the value it reports as the company's number of employees for FY2025, not as a different kind of figure or a figure for another period.
• The response shows its basis: it names the document it read the figure from and identifies it as the final FY2025 results summary, not the superseded draft the task warns about.
• The response commits to a single employee count for "employees": any employee count it gives as its answer in the prose matches the JSON value; a figure it explicitly attributes to the superseded draft does not count as an alternative.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the final FY2025 results summary could not be found or read, it does not present the value as confirmed.

Pre-registered rubric

• The "revenue" value in the final JSON object is the number 40126615 exactly (a JSON number, not a quoted string).
• The final reply includes a JSON object with the field "revenue" whose value is a bare JSON number written out in full — not a quoted string, a range, an abbreviated figure with a scale word or suffix, or text with a currency symbol, separators or units inside the value.
• The response labels the value it reports as the company's FY2025 revenue in US dollars, not as a different kind of figure, a figure in another currency, or a figure for another period.
• The response shows its basis: it names the document it read the figure from and identifies it as the final FY2025 results summary, not the superseded draft the task warns about.
• The response commits to a single revenue figure for "revenue": any revenue figure it gives as its answer in the prose matches the JSON value; a figure it explicitly attributes to the superseded draft does not count as an alternative.
• The response does not present a guessed, assumed or placeholder value as if it were read from the vault: if it reports that a step failed or that the final FY2025 results summary could not be found or read, it does not present the value as confirmed.