prompt large_stringclasses 957
values | response large_stringlengths 50 1.4k |
|---|---|
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Set an availability_rule for menu_item ID 305 so it is only available Monday to Friday from 11:00 to 15:00 local kitchen time and mark it as unavailable at all other times. | {"episode_summary": "The agent created active availability rules for menu item 305 on Monday-Friday 11:00-15:00 and then added inactive Saturday/Sunday rules, but it failed because it did not inspect or reconcile pre-existing active rules, left a conflicting active Saturday rule in place, and incorrectly used inactive ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For my existing project 'Loft Apartment – Evening Ambience', create a fixture_group named 'Living Room Accent Lights' that includes the TV console RGBW strip and any wall-wash panels in the Living Ro... | {"episode_summary": "The agent resolved the existing project by name, listed fixtures to identify the TV console RGBW strip and Living Room wall-wash panels, checked existing fixture groups after a create error, confirmed the target group already existed, verified its members via the API, and reported that the group co... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As an applicant organization, create a new application in draft status for the funding_round with ID 103 including project_summary 'After-school arts labs in three community centers', outcomes ['serv... | {"episode_summary": "The agent successfully created a draft grant application for funding round 103 with the required project summary, target population, and total requested amount of 55000. It then added the two specified outcomes and the three budget items using the returned application ID, leaving the application in... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Deactivate any intake_forms in my clinic that are in status 'active' but have an expiry_rules date earlier than today by updating their status to 'draft' and appending '[EXPIRED]' to their metadata t... | {"episode_summary": "The agent assumed clinic_id=1 and guessed the current date, then called a bulk expiry tool and interpreted an empty result as meaning no expired forms existed. It did not verify the clinic scope, confirm the actual current date, enumerate active intake forms, inspect expiry_rules, or apply the requ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Add a new item called 'Classic Cheeseburger' with short_label 'ChzBrg', base_price 8.99, tax_rate 0.0825, and tags ['burger','cheese'] to the 'Burgers' category on the dine_in menu_board for store ID... | {"episode_summary": "The agent successfully added the Classic Cheeseburger item to the Burgers category on the dine_in menu board for store 101 by resolving the menu board and category, creating the item with the required price, tax, and label, upserting and attaching the burger and cheese tags, and linking the item to... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: From my existing pending matches, accept (set status to 'active') any match where the other user's location is within 5 miles of 'Seattle, WA' and their lifestyle_flags include 'pets-dog', and block ... | {"episode_summary": "The agent listed pending matches, resolved Seattle, WA coordinates, evaluated each pending match's age, location, and lifestyle flags, blocked the under-25 match, and left the other pending because it lacked the exact pets-dog flag.", "episode_skill": "For pending-match triage, first fetch pending ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Adjust an existing partially_paid invoice for my cat 'Luna' by adding a new line_item 'Labwork recheck' quantity 1 at unit_price 65.00, recalculate tax at 7.5%, and update payment_status from 'partia... | {"episode_summary": "The agent correctly identified Luna and the Chase Sapphire payment method, but failed to locate the existing partially paid invoice because no list-invoices tool was available. Instead of using the available get_invoice tool with likely candidate IDs or otherwise probing for the invoice, it stopped... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As an applicant, create a new admissions_application for the 'MS Computer Science' program in the 'Fall 2026' academic_term at 'Northfield University' with status 'started', then attach it to my exis... | {"episode_summary": "The agent created a new admissions application for MS Computer Science at Northfield University for Fall 2026 with status started by resolving the current applicant profile, institution, program, and academic term IDs, then calling the applicant-specific create application tool and confirming the r... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: In class "Math 4A", approve all pending portfolio submissions tagged with the assignment name "Fractions Worksheet #2" and then post a short feedback comment "Nice work on simplifying fractions; chec... | {"episode_summary": "The agent located Math 4A by probing classroom IDs after list endpoints failed, identified the Fractions Worksheet #2 assignment, approved the three pending portfolio items, and posted the required feedback comment on each approved item.", "episode_skill": "Resolve the target classroom ID before ac... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new goal for the 'movement' wellness_area targeting a streak_length of 14 days for the habit 'Evening walk 20 minutes', and set its target_metric_type to 'streak_length' with target_value 14... | {"episode_summary": "The agent resolved the requested 'movement' wellness area to ID 3, found the habit 'Evening walk 20 minutes' with ID 6 by listing habits in that area, and created an active goal with target_metric_type 'streak_length', target_value 14, and streak_length 14.", "episode_skill": "When creating a goal ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For my existing family group 'Lopez Family', create a new occasion 'Christmas 2025' on 2025-12-25 with theme 'Cozy Family Christmas', attach it to the group so all members can see it, and duplicate m... | {"episode_summary": "The agent resolved the Lopez Family group to id 1, found the Christmas 2024 occasion id 3, created a new Christmas 2025 occasion id 11 on 2025-12-25 with theme Cozy Family Christmas and attached it to the group, listed wishlists for the 2024 occasion, and duplicated wishlist id 4 to the new occasio... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate a report of all denied Duo authentications for the past 7 days, including timestamp, username, access device IP address, and denial reason. | {"episode_summary": "The agent identified the dedicated denied-authentication report tool, called it for the default 7-day window, and returned a formatted report containing all denied events with timestamp, username, access device IP, and denial reason, plus summary counts.", "episode_skill": "For denied Duo authentic... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For the form 'Post-Purchase Satisfaction Survey', create a distribution of type 'email_campaign' configured to send invitations to the email list segment 'Recent Buyers - Last 30 Days', and include t... | {"episode_summary": "The agent located the 'Post-Purchase Satisfaction Survey' form by listing user forms, identified its ID as 1, and created an email_campaign distribution with the required segment, subject line, and body text. The created distribution was returned successfully with draft status.", "episode_skill": "... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For customer 'GreenLeaf Logistics', fetch all reservations with status 'returned' for the last 90 days and generate a summary of total_rental_days, total_mileage_used, and total_amount_invoiced, then... | {"episode_summary": "The episode failed because no interaction trajectory was provided, so the required steps for customer 'GreenLeaf Logistics'—fetching returned reservations from the last 90 days, calculating total_rental_days, total_mileage_used, and total_amount_invoiced, and creating invoice metadata 'INV-GL-2025Q... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: On the survey titled 'Market Research – New Feature X', enable question-level validation on the numeric text question 'What is your monthly budget for this type of tool (in USD)?' so that it only acc... | {"episode_summary": "The agent located the survey 'Market Research – New Feature X' by exact title, listed its questions, identified the numeric budget question, and verified it already had integer_range validation from 10 to 5000 with the requested error message, so it reported that no update was needed.", "episode_sk... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As a buyer, create a new RFQ named 'Q1 Maintenance Supplies' with desired delivery date '2026-02-15' including line items: 500 units of SKU 'GLOV-NITR-5MIL', 50 units of SKU 'SAF-HARNESS-01', and 10 ... | {"episode_summary": "The agent successfully created a buyer RFQ named 'Q1 Maintenance Supplies' with desired delivery date 2026-02-15, added the three requested SKU line items with correct quantities, resolved the two supplier names to supplier IDs, and sent the RFQ to both suppliers.", "episode_skill": "To create and ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As a brand user, create a new Fall 2025 catalog tagged with season 'Fall 2025' that includes three SKUs in the 'Outerwear' category with case-pack size 6, minimum order quantity 12, and set a custom ... | {"episode_summary": "The episode failed because the provided interaction trajectory is empty, so no verifiable actions were taken to create the Fall 2025 catalog, add the three Outerwear SKUs, or configure the Urban Outfitters NYC price list.", "episode_skill": "Before creating catalogs, SKUs, or price lists, verify th... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: On my taxable account, place a limit buy order for 10 shares of a US semiconductor stock (asset_class=stock, sector=Technology, risk_rating=High) with a limit_price exactly 5% below its latest quote'... | {"episode_summary": "...", "episode_skill": "..."} |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: From my individual brokerage account, transfer $1,000 from my linked Schwab Bank checking account into the brokerage cash balance and schedule this transfer to occur on the next business day. | {"episode_summary": "The agent scheduled a $1,000 transfer from Schwab Bank Checking to the Individual Brokerage account but failed by guessing the next business day from account timestamps and selecting 2026-09-15 instead of 2026-09-14, and by using ACH_IN for a linked internal bank-to-brokerage transfer.", "episode_s... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: At the Main Campus, list all lunch vendors operating today and return the one with the earliest available time_slot that has at least 5 remaining order capacities. | {"episode_summary": "The agent failed to complete the Main Campus lunch vendor task, likely by not reliably retrieving all vendors operating today or not correctly filtering for the earliest time slot with at least 5 remaining order capacities.", "episode_skill": "For vendor availability tasks, first retrieve all vendo... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Simulate a guest_check_out event for room '502' in the 'Riverside Hotel', then verify that the associated automation sets occupancy_status to 'vacant_dirty', applies the 'reset_scene' to all devices,... | {"episode_summary": "The agent failed to complete the guest_check_out simulation for room 502 at Riverside Hotel and did not verify that the automation set occupancy_status to vacant_dirty, applied reset_scene to all devices, and locked the door_lock device.", "episode_skill": "Do not treat a hotel checkout automation ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new group called 'Contractors-Q1' containing the users 'alice.contractor@example.com', 'bob.contractor@example.com', and 'charlie.contractor@example.com', and apply a policy to this group th... | {"episode_summary": "The agent created/located the Contractors-Q1 group after create_group returned 500, added the three contractor users by email, created a group-scoped MFA policy requiring MFA with no remembered devices, and assigned the policy to the group.", "episode_skill": "For group-scoped MFA enforcement, firs... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Record the outcome of an existing tour for contact 'GreenLeaf Retail Group' at unit 'Retail Bay 3' in 'Lakeside Retail Center' by updating the tour's outcome to 'interested' and adding a follow-up no... | {"episode_summary": "The agent resolved the named contact, property, and unit to their IDs, located the matching existing tour for GreenLeaf Retail Group at Retail Bay 3, and updated that tour's outcome to 'interested' with a note requesting a draft term sheet within 3 business days.", "episode_skill": "When recording ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Retrieve all boards I currently follow that focus on 'Developer Tools' (i.e., board topics containing 'dev tools' or 'programming'), and for each board follow the curator if I am not already followin... | {"episode_summary": "...", "episode_skill": "..."} |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Request a paginated list of threads in the 'DevOps' category that are currently 'unanswered' (no accepted answer) and have a score of at least 2, returning page 3 with a page size of 20 including thr... | {"episode_summary": "The agent attempted to retrieve page 3 of a paginated list of DevOps threads that are unanswered and have a score of at least 2, but failed to return the correct result set with the required fields.", "episode_skill": "When querying paginated thread lists, explicitly verify and encode all constrain... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate and return my annual giving summary for the last calendar year, including total donated amount, number of donations, a breakdown by campaign, and the earliest and latest donation date in tha... | {"episode_summary": "The agent identified the dedicated annual giving summary tool, assumed 2024 as the last calendar year, queried it, received zero donations, and returned a clear summary with total amount, donation count, earliest/latest dates, an empty campaign breakdown, and an offer to check another year.", "epis... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As a meal_provider managing location 'Downtown HQ', upload a new weekly menu for the week starting 2025-02-10 that includes three menu_items—'Grilled Tofu Bowl', 'Turkey Power Salad', and 'Lentil Veg... | {"episode_summary": "The agent resolved company, Downtown HQ location, and existing provider-location mapping, created a Monday-Friday menu for 2025-02-10 to 2025-02-14, added three menu items with nutrition, allergens, price, and tags, created lunch fulfillments for each item on each weekday, and retrieved the provide... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: In my existing classroom "Period 2 Algebra", configure the behavior settings so that "Completed Homework" gives +2 points, "Helping Others" gives +3 points, and a new negative behavior "Off Task" ded... | {"episode_summary": "The agent located the existing classroom 'Period 2 Algebra' by name, retrieved its behavior list after correcting an invalid string classroom_id to an integer, then updated 'Completed Homework' to +2 points, 'Helping Others' to +3 points, confirmed 'Off Task' already existed as a -1 negative behavi... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate a dashboard summary for Hospital ID 3 for the date 2026-01-02 showing overall bed occupancy_rate by ward (as percentages), the total number of scheduled vs completed procedures, and operatin... | {"episode_summary": "Agent identified the dedicated daily bed and OR dashboard tool, called it for hospital_id 3 and date 2026-01-02, and formatted the returned ward occupancy, procedure counts, and OR utilization into a clear dashboard summary.", "episode_skill": "For hospital daily operational dashboard requests, fir... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Retrieve the top 10 interviewers by number of completed interviews in the past 30 days across all departments, along with their total interviews count and average candidate recommendation_flag distri... | {"episode_summary": "The agent identified the dedicated top-interviewers report tool, called it with days=30 and limit=10, and returned the resulting 7 interviewers with completed interview counts and recommendation_flag distributions in a formatted table with brief insights.", "episode_skill": "For top-N aggregate rep... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: On the survey titled 'Market Research – New Feature X', enable question-level validation on the numeric text question 'What is your monthly budget for this type of tool (in USD)?' so that it only acc... | {"episode_summary": "The agent located the survey by exact title, listed its questions after correcting a string-to-integer survey_id error, identified the numeric budget question, and applied/confirmed integer_range validation from 10 to 5000 with the requested custom error message.", "episode_skill": "To configure qu... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Find all incidents of type 'damage' discovered between '2026-03-01T00:00:00Z' and '2026-03-31T23:59:59Z' for shipments where the mode is 'road' and return their shipment reference_numbers and inciden... | {"episode_summary": "The agent failed to complete the request to find damage incidents discovered in March 2026 for road-mode shipments and return their shipment reference numbers and incident severities.", "episode_skill": "When querying incidents tied to shipments, explicitly join incidents to shipments and apply all... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As the current subscriber, simulate a failed monthly renewal billing_event for my 'Manga Starter Basic' plan with amount 4.99 USD, currency 'USD', outcome_code 'card_declined', and then show my updat... | {"episode_summary": "The agent retrieved the subscriber profile, found no existing subscription, created a new active 'Manga Starter Basic' subscription, recorded a card_declined renewal billing event, and then reported the subscription as active with Shonen Trials access still granted. It failed because the failed ren... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Start a new application for the 'Children’s Comprehensive Coverage' program for both of my children in my household and pre-populate it with our current demographics, income, and residency informatio... | {"episode_summary": "The agent located the user's household, identified both children, found the Children's Comprehensive Coverage program, created an application after initial 500 errors, added both children as members, and populated residency. However, repeated 500 errors prevented income and employment fields from b... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As a patient, schedule a new video visit for a primary care consultation with any available provider at my home clinic between 6–8 PM local time next Wednesday using my Blue Cross PPO insurance, and ... | {"episode_summary": "The agent tried to schedule a telehealth primary care visit at the patient's home clinic for next Wednesday between 6-8 PM local time using Blue Cross PPO. It retrieved the patient, clinic, visit type, and insurance plan, but repeatedly failed to call the availability search due to a 422 tool error... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate a single-use booking link for my '15-Minute Quick Chat' event type that expires after the first successful booking and then return the URL of that single-use link. | {"episode_summary": "The agent found the correct '15-Minute Quick Chat' event type but repeatedly failed to create a new single-use link due to malformed tool-call arguments and repeated 422 errors. Instead of reporting the creation failure or fixing the call format, it listed existing single-use links and returned an ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For pharmacy branch_id 'PHX-003', update its allowed_services to include ['vaccines','compounding','med_sync'], set hours for 'Monday' to '08:00-19:00', and change contact_info.phone to '555-447-9000... | {"episode_summary": "The agent updated pharmacy branch PHX-003 by inspecting available tools, retrieving the current pharmacy record, and applying all requested configuration changes in a single patch_pharmacy call: allowed_services set to vaccines, compounding, and med_sync; Monday hours set to 08:00-19:00; and phone ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create an application_form for the program 'Downtown Diabetes Education – January 2026' that includes structured questions for full_name (string, required), date_of_birth (date, required), primary_la... | {"episode_summary": "The agent located program ID 335 for Downtown Diabetes Education – January 2026, attempted to create an application form, then queried the active form and found form 1001 already active with all requested fields, so it reported the existing form as complete.", "episode_skill": "For program applicat... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Discover 3 low-carb breakfast recipes under 400 calories per serving that do not contain eggs as an ingredient and return their ingredients lists and nutrition summaries. | {"episode_summary": "The agent first tried the dedicated low-carb no-egg breakfast discovery tool, but it returned a 500 error. It then fell back to the generic recipe search with breakfast meal type, a 400-calorie per-serving cap, exact exclusions for egg variants, and a limit of 3. After receiving three candidate rec... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Retrieve a list of all users who have not completed Duo enrollment within the last 14 days and return their usernames and email addresses. | {"episode_summary": "The agent identified the dedicated MCP report tool for users not enrolled within a specified number of days, called it with days=14, and returned the resulting usernames and email addresses in a table with a total count.", "episode_skill": "When asked for a time-window enrollment report, prefer a p... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: In the 'All-Day Diner Menu' for 'Sunrise Diner Downtown', create a menu_item called 'Avocado Toast' in the 'Starters' section with short_description 'Sourdough toast with smashed avocado and cherry t... | {"episode_summary": "...", "episode_skill": "..."} |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: In the 'Starters' menu_section, add a new menu_item named 'Crispy Cauliflower Bites' with the short description 'Fried cauliflower florets with chili-lime aioli', price 9.50, preparation_station 'Fry... | {"episode_summary": "The agent located the target location and the 'Starters' menu section, corrected an integer parameter error, reused existing vegan, spicy, and gluten-free tags, created the 'Crispy Cauliflower Bites' menu item with the required details, and attached the tags to the new item.", "episode_skill": "Res... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Request a paginated list of threads in the 'DevOps' category that are currently 'unanswered' (no accepted answer) and have a score of at least 2, returning page 3 with a page size of 20 including thr... | {"episode_summary": "The agent failed to return the requested paginated list of DevOps threads that are unanswered, have a score of at least 2, and are on page 3 with a page size of 20, likely due to incorrect filtering, pagination, or field selection.", "episode_skill": "When requesting a paginated thread list, explic... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Get the neighborhood information for the property at '890 Elm Ave, Long Beach, CA 90813', including walkability metrics, median home price in the neighborhood, and crime statistics if available, and ... | {"episode_summary": "The agent parsed the provided Long Beach address into address components, attempted the direct by-address neighborhood endpoint, recovered from a 422 error by resolving the property ID from the address, then fetched neighborhood info by property ID and returned an aggregated response with neighborh... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Update conversation ID 562 by adding tags ['Truck Buyer', 'Lead Source:Website', 'Towing Package Interest'], reassign it from the sales queue to the finance queue, and change its topic to 'finance' w... | {"episode_summary": "The agent successfully updated conversation 562 by adding the requested tags, resolving the finance queue ID, and updating the conversation's queue and topic to finance while leaving its status active.", "episode_skill": "For conversation updates involving tags, queue, and topic, first add tags wit... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For the task titled 'Implement homepage hero in React' on the 'Q2 Website Redesign' board, set a dependency so that it is blocked by the task 'Design new homepage hero section', and update the blocke... | {"episode_summary": "The agent found the Q2 Website Redesign board and both tasks, then correctly made 'Implement homepage hero in React' depend on 'Design new homepage hero section'. It failed by updating the description of the design/blocking task instead of the blocked implementation task, so the required descriptio... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new SLA policy called 'Standard Support SLA' that requires first response within 4 business hours and resolution within 48 business hours for tickets with priority 'normal' submitted via ema... | {"episode_summary": "The agent attempted to create the Standard Support SLA, received a 500, then listed SLA policies and found an existing policy with the required 4h/48h business-hour targets. It added priority=normal and email/web-form channel conditions, listed Pro customers and organizations, and confirmed the pol... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Find all MRI assets that are currently in status 'in service' in Building 'Imaging Center' and return their asset IDs, room locations, and next scheduled preventive maintenance dates. | {"episode_summary": "The agent searched for MRI assets with status 'in service' in the Imaging Center, retrieved three matching assets with their room locations, then queried preventive maintenance schedules for each asset ID and returned the asset IDs, room locations, and next scheduled PM dates.", "episode_skill": "F... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a monthly subscription for client 'Acme Design Studio' named 'Ongoing support retainer' that generates an invoice every month on the 1st with frequency 'monthly', next_issue_date '2025-02-01',... | {"episode_summary": "The agent located the client, product, and tax rate, but did not verify the client's default payment method. After a validation error when passing payment_method_id as null, it created a new auto-billing subscription without a payment method, producing a duplicate/incorrect subscription instead of ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Retrieve all open orders originating from the 'delivery' channel that have a promised_time in the next 30 minutes, and for each order that is not yet started, update the fire_time on their associated... | {"episode_summary": "The agent listed open delivery orders, inferred the current time from timestamps, selected the order with promised_time within 30 minutes, retrieved its queued tickets, and updated both ticket fire_times to an earlier time to prioritize them over dine-in tickets.", "episode_skill": "For urgent orde... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: On the question with ID 68420317, upvote the answer that currently has the highest score, then add a comment saying 'This solution worked for me on Node.js 18 with only minor changes, thanks!'. | {"episode_summary": "The agent successfully completed the task by retrieving answers for question 68420317 sorted by score, identifying answer 2401 as the highest-scoring answer, upvoting it, and then adding the requested comment to that same answer.", "episode_skill": "When asked to upvote and comment on the highest-s... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Apply a new coupon code 'WELCOME20' that gives a 20% discount for the first 3 billing cycles to the active subscription belonging to customer email 'jane.doe@example.com'. | {"episode_summary": "The agent located customer jane.doe@example.com, selected their active subscription, confirmed the WELCOME20 coupon existed with a 20% discount for 3 cycles, applied it to the subscription, and verified the subscription_coupons record showed remaining_cycles=3.", "episode_skill": "To apply a coupon... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate a personalized feed of boards for me by returning the 20 most recently updated public boards created by users I follow or that are tagged with any of my top 5 most-followed tags, including f... | {"episode_summary": "The agent identified that the request matched the dedicated personalized board feed tool, called mcp_tool_mcp_server_get_personalized_board_feed with limit=20, received 7 available boards containing the required fields, and returned them in a clear table with a summary and optional follow-up action... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Filter the TV shows catalog to only display comedies with at least 4-star user ratings and return the first 10 results sorted by popularity. | {"episode_summary": "The agent identified the catalog search tool, called list_titles with genre=Comedy, min_avg_rating=4, sort_by=popularity, sort_order=desc, and limit=10, then returned the 3 matching titles in a table and noted that fewer than 10 results existed.", "episode_skill": "For catalog filtering requests, t... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a work_order for vehicle asset_id 'TRUCK-145' with vendor 'QuickLube Downtown', service_date '2024-09-20', odometer_at_service 86500, including line_items for labor 'Oil and filter change' at ... | {"episode_summary": "The agent created a work order for vehicle TRUCK-145 by resolving the vehicle to ID 1 and finding the existing vendor QuickLube Downtown with ID 2. It created work order 11 with service date 2024-09-20 and odometer 86500, added a labor item for $80, added two parts items totaling $57, and updated t... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As a clinician, for the completed visit with encounter ID 98765, create an encounter summary noting diagnosis 'acute sinusitis', recommended follow-up 'telehealth visit in 2 weeks if symptoms persist... | {"episode_summary": "The agent attempted to create an encounter summary for encounter 98765 with diagnosis acute sinusitis, follow-up telehealth in 2 weeks if symptoms persist, and a brief clinical note. It repeatedly called the summary creation endpoint and encounter lookup, received 500 errors and one validation erro... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Update my default payment_method token to the one ending in '7843' and then place a dinner order at 'Parkside Food Truck' today at 7:15–7:30 PM for 3x 'Veggie Tacos' (note: 'no cheese') and 1x 'Spark... | {"episode_summary": "The agent set the payment method ending in 7843 as default and found the vendor and menu items, but failed to determine the correct current date. It guessed multiple service dates, eventually used 2026-07-15 inferred from record timestamps, selected a 19:00-19:00 timeslot with zero capacity, and cr... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate a predictive maintenance summary via the reporting API that lists, for the tenant 'Acme Manufacturing', all devices whose vibration trend over the last 30 days exceeds a 20% increase compare... | {"episode_summary": "...", "episode_skill": "..."} |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As a scheduler for a clinician account, update Dr. Lee's availability so that new 20-minute telehealth slots are created every Monday, Wednesday, and Friday from 9:00am to 12:00pm for visit types 'ur... | {"episode_summary": "The agent located Dr. Susan Lee and the required visit types, then created three availability rules for Monday, Wednesday, and Friday from 09:00 to 12:00 with 20-minute slots for urgent care and medication refill. It used an assumed current date and did not call the slot-generation tool or verify t... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For the task titled 'Implement homepage hero in React' on the 'Q2 Website Redesign' board, set a dependency so that it is blocked by the task 'Design new homepage hero section', and update the blocke... | {"episode_summary": "The agent correctly located the 'Q2 Website Redesign' board, identified the two tasks, and added the dependency so 'Implement homepage hero in React' is blocked by 'Design new homepage hero section'. However, it updated the description of the blocking task ('Design new homepage hero section') inste... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Set up a webhook subscription that sends a POST request to https://api.mychannelmanager.com/urbanstay-webhook for any reservation status changes (created, modified, canceled) for my properties in Ams... | {"episode_summary": "The agent resolved Amsterdam's city_id using the city list tool, then created an active webhook subscription for reservation.status.change events scoped to that city and pointing to the requested target URL. The returned subscription confirmed the setup for reservation status changes in Amsterdam."... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a weekly recurring subscription for myself at the 'Downtown HQ' location that orders the menu_item with ID 5120 every Monday, Wednesday, and Friday for lunch starting on 2025-02-03, and apply ... | {"episode_summary": "The agent resolved the user's company and Downtown HQ location ID, checked the dietary profile, computed the Mon/Wed/Fri bitmask as 21, and created an active weekly lunch subscription for menu item 5120 starting 2025-02-03 with dietary validation enabled; validation passed with no violations.", "ep... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new service request for a 'Standard Windows 11 Laptop' for user jane.smith@example.com with needed-by date '2025-01-15', cost center 'IT-ENG', requested for the 'Remote Work' bundle (laptop ... | {"episode_summary": "The agent created a service request for a Standard Windows 11 Laptop for jane.smith@example.com by calling the service request creation tool with the needed-by date, cost center, Remote Work bundle, and Hardware Fulfillment assignment group. The request was successfully created as REQ0010002.", "ep... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a limited-time event named 'Winter Frost Fest' that runs from December 20 to January 5, associates three exclusive cosmetic items ('Snowflake Aura', 'Penguin Hoodie', 'Ice Shard Trail') with l... | {"episode_summary": "The agent attempted to create the Winter Frost Fest event and its associated legendary cosmetics and shop entries, but write tools repeatedly returned 500 errors. It then used read-only list endpoints to verify that the event, the three legendary cosmetics, the event-cosmetic associations, and the ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For the existing event 'Park Cleanup – Riverside Trail' next Saturday, bulk-assign all volunteers tagged 'driver' who are available between 8:00 AM and 12:00 PM to the 'Trash Pickup Lead' role until ... | {"episode_summary": "The agent failed to complete the bulk assignment for the Park Cleanup – Riverside Trail event, not reliably identifying all 'driver' volunteers available 8:00 AM–12:00 PM next Saturday, filling the 'Trash Pickup Lead' role to capacity 6, and placing remaining matches on the waitlist.", "episode_ski... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a 4-week study plan in table form to prepare a college student for an introductory calculus exam, including daily topics, suggested practice problems, and short review prompts for spaced repet... | {"episode_summary": "The agent recognized that the request was a content-generation task requiring no external tools and directly produced a 4-week introductory calculus study plan in table form. It included daily topics, suggested practice problems, spaced repetition review prompts, weekly review days, a weekly focus ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new message template with ID 'greeting_sales_01' for the sales team whose body says 'Hi, this is {{agent_name}} with {{dealer_name}}; what vehicle are you interested in today?' and mark it a... | {"episode_summary": "The agent resolved the sales team to team_id 1, then attempted to create the requested greeting template multiple times. All create_message_template calls returned 500 errors, so the agent stopped and reported a likely server-side issue with recommended next steps.", "episode_skill": "To create a t... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate a summary report of all my bookings as an owner in Q2 2026, including total number of approved bookings, total rental days, total base revenue, total penalty amounts applied (e.g., late_retu... | {"episode_summary": "Agent attempted to generate a Q2 2026 owner bookings summary but queried the owner bookings summary tool with 2026-06-01 to 2026-08-31, misinterpreting Q2 as June-August. It then presented the returned metrics as Q2 2026 and claimed coverage of April 1 through June 30 without re-querying the correc... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Run a what-if simulation on the 'Fall 2025 Dinner' menu to calculate the new overall menu food cost percentage if all mains have their prices increased by 8% while ingredient costs remain constant, a... | {"episode_summary": "The agent resolved the 'Fall 2025 Dinner' menu, attempted to retrieve menu pricing and item data, encountered repeated 500 errors on menu-item endpoints, then pivoted to checking existing menu cost summaries. It found a matching what-if summary for an 8% mains price increase and returned the origin... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As an organization admin, create a new course_catalog named '2025 Onboarding Essentials' scoped to categories 'Compliance', 'Product Training', and 'Workplace Safety', then add 15 existing courses in... | {"episode_summary": "The agent failed to complete the task of creating the '2025 Onboarding Essentials' course catalog scoped to Compliance, Product Training, and Workplace Safety, and adding 15 existing courses from those categories. The failure appears to stem from not reliably confirming the created catalog, correct... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Using the Okta API Access Management APIs, create a custom authorization server named 'payments-api', define a custom scope 'transactions:write', and configure a policy that allows only clients with ... | {"episode_summary": "The agent identified the Okta API Access Management tools needed to create the payments-api authorization server, transactions:write scope, and client-restricted policy. It repeatedly attempted create_authz_server with different ID formats, but the tool returned persistent 500 errors. The agent the... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new concept called 'Downtown Bistro' and under it create a dinner menu named 'Fall 2025 Dinner' with categories 'Starters', 'Mains', and 'Desserts', each with an availability window of 5pm–1... | {"episode_summary": "The agent failed to create the 'Downtown Bistro' concept and 'Fall 2025 Dinner' menu with the required Starters, Mains, and Desserts categories and 5pm–10pm daily availability from 2025-09-15 to 2025-12-15.", "episode_skill": "When creating menu structures, explicitly create the parent concept, men... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For my SaaS product 'Acme CRM', create a new public feedback_board named 'Feature Requests' with posting_rules that allow any authenticated user to submit idea-type feedback_posts and set its descrip... | {"episode_summary": "The agent looked up the Acme CRM product, then repeatedly called create_feedback_board with guessed posting_rules JSON formats and received 500 errors. After the failures, it listed existing boards, found a pre-existing 'Feature Requests' board matching the requested settings, and incorrectly repor... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: On job '#J-31567' for customer 'Smith Family HVAC', add three line items to an estimate pulled from the pricebook (1x 'AC Tune-Up', 1x 'MERV 11 Filter 20x25x1', and 1x 'Surge Protector – HVAC'), then... | {"episode_summary": "The agent resolved customer 'Smith Family HVAC' and job '#J-31567', looked up the three requested pricebook items, created estimate EST-31567, and added the three line items with the correct quantities and prices. Recalculation attempts failed with validation errors, so the agent converted the esti... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As an instructor, generate a revenue_report for my instructor account summarizing total enrollments, refunds, and net revenue per month for the last 6 complete months across all my courses, and retur... | {"episode_summary": "...", "episode_skill": "..."} |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new beneficiary_address tagged 'Binance Main Account' for BTC with a destination_tag 'whitelisted_exchange' and then initiate an external_transfer of 1.2 BTC from 'Trading Hot Wallet' to thi... | {"episode_summary": "The agent created the requested beneficiary tag, beneficiary address, tag association, and external transfer, but failed by using an invalid fabricated BTC address containing Cyrillic text instead of stopping to request a valid external address. It then proceeded to initiate a 1.2 BTC external tran... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As an applicant organization, create a new application in draft status for the funding_round with ID 103 including project_summary 'After-school arts labs in three community centers', outcomes ['serv... | {"episode_summary": "The agent created a draft grant application for funding round 103 using the provided project summary, target population, and a computed total requested amount of 55000. It then used the returned application ID to add two outcomes and three budget items, completing the requested draft application wi... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For international order 'EBAY-44771' shipping from the United States to Germany with 2 units of SKU 'POSTER-A3' (weight 0.2 kg each, HS code 491191, unit value 15 EUR), create a customs declaration l... | {"episode_summary": "Agent resolved order EBAY-44771, verified the POSTER-A3 line item and existing international shipment, then attempted to create the required customs declaration. After repeated 500 errors, it created the customs line item against an existing declaration ID and reported that the declaration endpoint... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new shipment with origin 'Chicago, IL', destination 'Los Angeles, CA', mode 'road', reference_numbers ['PO-98421', 'BOL-5511'], planned_pickup_date '2026-01-10', and planned_delivery_date '2... | {"episode_summary": "The agent successfully created a road shipment from Chicago, IL to Los Angeles, CA with planned pickup and delivery dates, attached the PO and BOL reference numbers, resolved the carrier names to organization IDs, and created two sequential shipment legs with the correct carriers and dates.", "epis... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Create a new fundraising campaign titled '2025 Education Fund' with a $50,000 goal, suggested donation amounts of $25, $50, $100, and $250, set the default currency to USD, and enable both one-time a... | {"episode_summary": "The agent was tasked with creating a '2025 Education Fund' campaign with a $50,000 goal, suggested donations of $25/$50/$100/$250, USD currency, and one-time plus monthly recurring donations, but the episode ended in failure with no completed or verified campaign creation.", "episode_skill": "When ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Set up a one-time payout of $2,500 USD to a connected account with ID 'acct_98765', to be paid out to their default bank account in USD, and include the statement descriptor 'Marketplace Payout March... | {"episode_summary": "The agent tried to create a one-time $2,500 USD payout to connected account acct_98765 with statement descriptor 'Marketplace Payout March'. It initially sent an invalid payout request using the wrong parameter name 'connection_account_id' instead of 'connected_account_id', then created a payout bu... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For all active jobs scheduled next Monday for workers 'John D.' and 'Maria L.' within the city of Denver, generate an optimized daily route that orders their visits to minimize estimated_drive_time, ... | {"episode_summary": "...", "episode_skill": "..."} |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Subscribe me to the channel with channel_id 88 using a monthly subscription plan and return my updated subscription status and renewal date. | {"episode_summary": "The agent listed subscription plans for channel 88, selected the monthly standard plan, created an active subscription with a start date and one-month renewal date, and returned the updated active status and renewal date.", "episode_skill": "For channel subscription requests, first list the target ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Retrieve the current air quality index (AQI) and detailed pollutant breakdown (PM2.5, PM10, O3, NO2) for Los Angeles, California. | {"episode_summary": "The agent retrieved Los Angeles, California air quality by first searching for the location, relaxing search filters after an empty result, obtaining location_id 4, then calling the current air quality tool and reporting AQI 110 with PM2.5, PM10, O3, NO2, SO2, CO, and observation time.", "episode_s... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Cancel an existing reservation for the 'LEGO Architecture Skyline' item in my 'Gadgets & Hobbies' registry by updating its reservation_status from 'reserved' to 'cancelled' and decreasing the reserve... | {"episode_summary": "The agent located the 'Gadgets & Hobbies' registry under the 30th Birthday occasion, found the 'LEGO Architecture Skyline' item, listed its active reservation, and cancelled reservation 1 with adjust_item_quantities=true, reducing the item's reserved quantity from 1 to 0.", "episode_skill": "To can... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Design a double-sided business card for 'Mia Johnson, Freelance UX Designer' where the front side includes her name, title, email (mia@uxmia.design), website (www.uxmia.design), and phone number (+1 ... | {"episode_summary": "The agent created a 1050x600 business_card document, added a front page and a back page, then built the front with centered text elements for name, title, email, website, and phone, and the back with a centered circular shape logo placeholder and centered tagline, using base element creation follow... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For my existing proposal with ID 87, update the budget by adding three line_items: (category 'personnel', amount 95000, justification_short 'PI and postdoc salary support'), (category 'equipment', am... | {"episode_summary": "The agent successfully updated proposal 87 by adding three budget line items in a single batch call: personnel for $95,000, equipment for $40,000, and travel for $15,000, each with the requested justification. The tool returned the created line items, confirming the budget update.", "episode_skill"... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For the venue 'Library Food Court' on 'Lakeside Campus', use the API to create an order with one 'Grilled Chicken Wrap' (size: large from the size option_set) and one 'Iced Coffee' (with 'oat milk' o... | {"episode_summary": "The agent resolved the campus, venue, menu items, option items, and earliest non-full pickup window, but initially called create_order without required quantity fields. After retrying with quantity, the order was created with total_amount_cents 1250 and empty options arrays, meaning the large wrap ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Start a charging_session for vehicle VIN '1N4BZ0CP1DC406001' on connector 1 of the 'ABB Terra 54' charger at 'Oakland Depot A' with start_time '2025-01-10T08:30:00-08:00' and start_soc 20%, then end ... | {"episode_summary": "The agent resolved the requested site, vehicle VIN, charger, connector, and tariff to internal IDs, started charging session 7 with the specified start time and 20% SOC, retrieved the tariff pricing rules, calculated 40.3 kWh from the vehicle's 62 kWh battery and 20% to 85% SOC change, applied the ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Configure an automation rule on the project 'Quarterly NPS Q1 2026' so that whenever a response is submitted with an NPS score from 0 to 6, a follow-up task is created tagged 'churn risk', assigned t... | {"episode_summary": "The agent identified the organization and project 'Quarterly NPS Q1 2026', confirmed the existing 'churn risk' tag and 'Customer Success - Team A' queue, and created an active response_submitted automation rule that triggers on NPS scores 0-6 and creates an open follow-up task in that queue with th... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Update customer profile with ID 'CUST-2045' by adding a new payment method token 'PMT-TOK-7821', uploading driver license metadata with number 'D1234567' expiring on '2028-06-30', and setting their v... | {"episode_summary": "The agent updated customer CUST-2045 by adding driver license D1234567 expiring 2028-06-30 and setting verification_status to verified. The payment method token PMT-TOK-7821 could not be added due to repeated 500 errors, so the agent reported partial completion.", "episode_skill": "For multi-part p... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Flag an existing review with ID 'rev_98765' for 'ACME Retail' as inappropriate due to 'harassment or hate speech', add a reporter note saying 'Contains personal insults and discriminatory language', ... | {"episode_summary": "The agent resolved ACME Retail to company_id 9, then created a review moderation flag for rev_98765 with reason 'harassment or hate speech', reporter note, and moderation_status 'pending review', completing the moderation queue update.", "episode_skill": "To flag a company review, first search the ... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Apply using my default resume and a generic cover letter that says 'I am excited to apply for this role and believe my experience in Python and SQL makes me a strong fit' to all 'junior software engi... | {"episode_summary": "The agent searched for junior software engineer jobs in Austin, TX with salaries over $80,000 and direct application enabled, relaxed an over-specific experience-level filter after an empty result, selected the first 3 matching jobs, retrieved the default resume and existing default cover letter wi... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For my course 'Financial Markets', reschedule all upcoming assignment deadlines by postponing each due date by exactly 3 days while preserving the time of day, and return the updated deadlines. | {"episode_summary": "Agent called the bulk reschedule tool for 'Financial Markets' with a 3-day offset, received an empty updated_assignments list, and reported that no upcoming assignments were found without verifying the assignment list or retrieving updated due dates, so the requested updated deadlines were not retu... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For my match with match_id 'MATCH-3012', mark the milestone whose goal_checkpoint is 'Shadow mentor in two stakeholder meetings' as completed by setting its completion_flag to true and recording the ... | {"episode_summary": "The agent identified the correct milestone-update tool and marked the milestone for match MATCH-3012 with goal_checkpoint 'Shadow mentor in two stakeholder meetings' as completed, allowing the system to record the completion timestamp automatically.", "episode_skill": "When a milestone must be comp... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: As a dispatcher, create a trip for ride_request ID 'req_5678' by assigning driver ID 'drv_221', vehicle ID 'veh_77', setting planned_pickup_time to '2025-01-15T08:45:00', planned_dropoff_time to '202... | {"episode_summary": "The agent created a scheduled trip for ride request req_5678 by first listing ride requests to resolve the missing required rider_id (rider_2002), then calling create_trip with the requested driver, vehicle, pickup/dropoff times, and status, and finally reporting the created trip details.", "episod... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: For all confirmed daycare bookings on 2025-03-05, assign each dog over 40 lbs to a 'Large Dog Playgroup' and each dog with temperament_tag 'shy' to a 'Calm Playgroup', ensuring no playgroup exceeds i... | {"episode_summary": "The agent failed to complete the daycare playgroup assignment task for confirmed 2025-03-05 bookings, without successfully filtering confirmed bookings, validating playgroup capacities, or updating assignments for dogs over 40 lbs and shy dogs.", "episode_skill": "Before assigning playgroups, restr... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate a summary report of all referrals sent from my organization 'Greenfield Animal Clinic' during December 2025, including for each referral the case ID, patient name, requested specialty, recei... | {"episode_summary": "The agent identified the dedicated report tool for sent referrals, queried it for 'Greenfield Animal Clinic' from 2025-12-01 through 2025-12-31, received five referrals with the required fields, and formatted a summary report including a table, key statistics, and a note for the referral with no st... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Assign the currently published 'Data Protection Policy' to all employees in the 'Finance' and 'HR' departments with due_date '2026-01-31', required_acknowledgment set to true, and initial completion_... | {"episode_summary": "The agent found the Data Protection Policy and two published versions, then selected version 1.5 (id 1289) based on a later effective date instead of confirming the currently published/latest version. It bulk-assigned that version to Finance and HR with the requested due date, acknowledgment, and s... |
Analyze the task below and return a concise reusable policy skill as JSON with episode_summary and episode_skill.
Task: Generate personalized video recommendations for my account based on my watch history but exclude any videos from channels I am already subscribed to and return the top 10 recommendations. | {"episode_summary": "...", "episode_skill": "..."} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.