▸case-01 Can you inspect my vault's concept notes to identify potential duplicate topics that ought to be combined? Please provide a summary object containing the total scanned count along with an array of candidate pairings specifying page 'a', page 'b', and a confidence rating. | fail→fail | 16,408 | 15,006 | -9% | 1 | 1 | 0% | 1,721 | 502 | -71% | 0 | 0 | — |
▸case-19 In my architecture vault, concept note A is titled 'Load-Bearing Wall' and note B is titled 'Structural Wall'. Their titles differ, but their definitions describe the same structural function. How should these two notes be processed? | fail→fail | 10,997 | 9,838 | -11% | 1 | 1 | 0% | 1,921 | 1,004 | -48% | 0 | 0 | — |
▸case-02 I need to clean up redundant pages in my knowledge base. Please run a check across concept pages to locate entries that seem to point to the same entity. Output the result in structured JSON showing the number of scanned items and the detected candidate pairs with a confidence score for each match. | fail→fail | 17,260 | 10,410 | -40% | 1 | 1 | 0% | 2,411 | 500 | -79% | 0 | 0 | — |
▸case-03 Check my vault for overlapping concept documents that should probably be consolidated into single entries. Return a JSON structure that reports the total scanned count and lists candidate pairs (a and b) alongside a confidence value. | fail→fail | 9,729 | 6,271 | -36% | 1 | 1 | 0% | 1,556 | 465 | -70% | 0 | 0 | — |
▸case-04 I'm looking for potential duplicate entries in my Obsidian vault's machine learning notes. Perform a search across the notes and flag candidate pairs if their search similarity metric is above 5.0. Output the findings in a JSON summary showing scanned note count and candidate pairs. | fail→fail | 12,750 | 14,919 | +17% | 1 | 1 | 0% | 2,566 | 539 | -79% | 0 | 0 | — |
▸case-05 I only want you to scan 3 concept notes in my research vault— specifically 'Neural Networks', 'Deep Learning', and 'Perceptrons'— to check if any should be merged. Return a structured JSON summary with the total scanned count and matching pairs. | fail→fail | 10,948 | 15,026 | +37% | 1 | 1 | 0% | 1,850 | 1,851 | +0% | 0 | 0 | — |
▸case-06 Check my vault's philosophy concept pages for redundant topics that could be merged. When looking for potential matches for a concept like 'Epistemology', should you search only for the exact title, or include related terms? Output the result in JSON format with scanned count and candidate matches. | pass→fail | 16,463 | 15,448 | -6% | 1 | 1 | 0% | 2,078 | 496 | -76% | 0 | 0 | — |
▸case-07 Audit my engineering concept pages for duplicate topics. Return a JSON report containing the list of potential duplicates under the key `matches` and total pages checked under `total_scanned`, with each match pair using `first_page` and `second_page`. | fail→fail | 12,882 | 5,599 | -57% | 1 | 1 | 0% | 1,277 | 485 | -62% | 0 | 0 | — |
▸case-08 When comparing vault search results in my medical notes to decide whether two concepts should be recommended for merging, how do you verify if two high-scoring search results actually refer to the same thing? Return a JSON breakdown of your findings for the vault. | fail→fail | 19,613 | 11,918 | -39% | 1 | 1 | 0% | 2,518 | 547 | -78% | 0 | 0 | — |
▸case-09 I want to find duplicate concept pages in my history notes vault. Should I copy and paste all my concept definitions directly into this chat for you to read, or can you inspect them yourself? Return JSON with scanned count and duplicate candidates. | fail→fail | 9,450 | 6,443 | -32% | 1 | 1 | 0% | 798 | 600 | -25% | 0 | 0 | — |
▸case-10 Scan my biology notes vault for concept pages that ought to be merged. Format the result as JSON where each pair in the match array includes `page1`, `page2`, and a `score_percentage` rating from 0 to 100. | fail→fail | 9,509 | 16,670 | +75% | 1 | 1 | 0% | 814 | 332 | -59% | 0 | 0 | — |
▸case-11 I want to identify duplicate concept pages in my physics vault. Should you run a raw terminal `grep -rn 'gravity' .` command across my markdown files to locate duplicates, or use a specialized vault search tool? Provide the output as JSON listing scanned count and candidate pairs. | fail→fail | 15,322 | 8,853 | -42% | 1 | 1 | 0% | 1,490 | 1,730 | +16% | 0 | 0 | — |
▸case-12 During a vault concept scan, if a search return yields a related concept page with a score of 8.5, what is the next step before adding it to the final merge recommendation list? Output your scan summary as JSON. | fail→pass | 11,692 | 9,826 | -16% | 1 | 1 | 0% | 1,076 | 991 | -8% | 0 | 0 | — |
▸case-13 My personal vault only contains 5 concept notes in total. I want to check for potential duplicates among them. How many concepts should be scanned, and what structured JSON shape should be returned? | fail→pass | 12,355 | 7,833 | -37% | 1 | 1 | 0% | 1,430 | 711 | -50% | 0 | 0 | — |
▸case-14 When searching my chemistry vault for candidate duplicates of a page titled 'Sodium Chloride', what query terms should be supplied? Return a JSON output reporting scanned count and candidate pairs. | fail→pass | 11,705 | 6,681 | -43% | 1 | 1 | 0% | 1,421 | 1,353 | -5% | 0 | 0 | — |
▸case-15 Scan my economics concept notes for duplicates and return a JSON object. I need to make sure each recommended merge item in the candidate list has a numeric metric indicating how certain the match is. What property name should store this value? | fail→fail | 7,553 | 9,695 | +28% | 1 | 1 | 0% | 1,370 | 1,028 | -25% | 0 | 0 | — |
▸case-16 Inspect my computer science vault for overlapping concepts like 'Async/Await' and 'Promises'. Output the consolidation report as a single JSON object with the required top-level count metric and array of candidate pairs. | pass→fail | 9,220 | 24,459 | +165% | 1 | 1 | 0% | 1,729 | 498 | -71% | 0 | 0 | — |
▸case-17 Suppose a vault search returns a concept match with a similarity score of 6.2 when scanning my literature vault. Should this concept pair be flagged as a merge candidate or excluded? Return JSON summary with scanned count and candidates. | pass→pass | 10,130 | 9,004 | -11% | 1 | 1 | 0% | 1,679 | 883 | -47% | 0 | 0 | — |
▸case-18 I need an automated scan of my mathematics concept notes to detect duplicate topics. Do not prompt me for confirmation between notes; complete the full batch scan across all concepts and output the structured JSON. | fail→fail | 2,686 | 9,689 | +261% | 1 | 1 | 0% | 365 | 438 | +20% | 0 | 0 | — |
▸case-20 I have confirmed that the concept note 'JS Promises' in my developer vault is a complete duplicate of 'JavaScript Promises'. Please move 'JS Promises.md' into my archive folder `/Archive/Concepts/` and remove its tag `#active-concept`. | pass→fail | 9,811 | 16,258 | +66% | 1 | 1 | 0% | 841 | 602 | -28% | 0 | 0 | — |
▸case-21 I want to merge the text content of two notes: 'Quantum Spooky Action' and 'Quantum Entanglement'. Please write a combined, cohesive markdown summary paragraph synthesizing their core ideas for my physics notes. | pass→fail | 11,048 | 17,157 | +55% | 1 | 1 | 0% | 880 | 521 | -41% | 0 | 0 | — |
▸case-22 My notes vault has over 50 concept pages with messy tags. Can you propose a hierarchical tagging taxonomy (e.g. `#science/physics/quantum`) to categorize these concepts cleanly? | pass→pass | 20,961 | 19,914 | -5% | 1 | 1 | 0% | 2,990 | 3,584 | +20% | 0 | 0 | — |