[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-braintrust-braintrust-design-human-eval-review":3,"mdc--g8iu91-key":38,"related-org-braintrust-braintrust-design-human-eval-review":329,"related-repo-braintrust-braintrust-design-human-eval-review":496},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":27,"repoUrl":28,"updatedAt":29,"license":30,"forks":31,"topics":32,"repo":33,"sourceUrl":36,"mdContent":37},"braintrust-design-human-eval-review","design human evaluation and review workflows","Design or audit human evaluation workflows and golden datasets for LLM applications and agents. Use to set up expert review, select review cases, write reviewer instructions, assign raters, capture rationales and confidence, measure inter-rater agreement with kappa or alpha, adjudicate disagreements, and preserve reviewed examples with provenance as a versioned reference set. Do not use to elicit the criteria or rubric in the first place, to validate a scorer once reference labels exist, or to implement the scorer.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"braintrust","Braintrust","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fbraintrust.png","braintrustdata",[13,17,20,23,26],{"name":14,"slug":15,"type":16},"Datasets","datasets","tag",{"name":18,"slug":19,"type":16},"LLM","llm",{"name":21,"slug":22,"type":16},"Evals","evals",{"name":24,"slug":25,"type":16},"Agents","agents",{"name":9,"slug":8,"type":16},7,"https:\u002F\u002Fgithub.com\u002Fbraintrustdata\u002Feval-library","2026-08-20T03:53:36.900129",null,0,[],{"repoUrl":28,"stars":27,"forks":31,"topics":34,"description":35},[],"Braintrust eval skills library","https:\u002F\u002Fgithub.com\u002Fbraintrustdata\u002Feval-library\u002Ftree\u002FHEAD\u002Fskills\u002Fbraintrust-design-human-eval-review","---\nname: braintrust-design-human-eval-review\ndescription: >-\n  Design or audit human evaluation workflows and golden datasets for LLM applications and agents.\n  Use to set up expert review, select review cases, write reviewer instructions, assign raters,\n  capture rationales and confidence, measure inter-rater agreement with kappa or alpha,\n  adjudicate disagreements, and preserve reviewed examples with provenance as a versioned\n  reference set. Do not use to elicit the criteria or rubric in the first place, to validate a\n  scorer once reference labels exist, or to implement the scorer.\n---\n\n# Design human review and the golden set\n\nContract: `references\u002Finteraction-contract.md`. Calibration, templates, provenance: `references\u002Freview-workflow.md`.\n\n## Trigger\n\n- \"Set up human review.\" \u002F \"Build a golden dataset.\" \u002F \"How should we adjudicate?\"\n- A rubric that exists and now needs applying at scale by people.\n- A judge that needs an anchor before it can be trusted.\n\n## Do\n\n1. Identify the expertise required and the decision the labels will support. Draft the review\n   form **before** asking about reviewer count.\n2. If no rubric or criteria exist yet, stop and elicit them from experts first. This skill\n   applies criteria; it does not invent them.\n3. Select cases deliberately: representative items, plus the ambiguous region where systems\n   differ, plus **every severe failure class**.\n4. Collect a short **rationale and confidence** with every label — the rationales are where the\n   rubric gets sharp.\n5. Have raters judge **independently before conferring**, then adjudicate on the record. A panel\n   deferring to whoever speaks first discards the benefit of a panel.\n6. Report agreement and adjudicate before the set calibrates anything, then preserve it as a\n   versioned dataset with per-item provenance and a refresh trigger.\n\n## Avoid\n\n- Do not treat majority opinion as ground truth without the relevant expertise.\n- Do not force adjudication over genuine construct ambiguity — record it; it is a finding about\n  the rubric.\n- Do not review only the easy middle; a golden set with no severe failures cannot validate a\n  safety scorer.\n- Do not compare a scorer against these labels here.\n\n## Check\n\n- Reviewer qualifications stated and matched to the construct.\n- Independent labeling before discussion; agreement computed, not assumed.\n- Adjudication decisions recorded with reasons.\n- Coverage spans clear successes, clear failures, ambiguous edges, severe failures.\n- Provenance per item; set versioned; refresh trigger defined.\n\n## Risk\n\n- Shared misunderstanding among reviewers produces a consistently wrong reference set that every\n  downstream number inherits — **high agreement is not validity**.\n- An unrepresentative review sample calibrates the scorer to the wrong distribution.\n- Ceiling effects erase the benefit of multiple raters: if nearly every item is obvious,\n  agreement is high and uninformative.\n- Reviewer fatigue degrades labels across a long session, invisibly without timestamps.\n\n## Braintrust\n\nRun review in **human review queues** with **per-criterion scoring and free-text notes**.\nPer-criterion is not optional: a single overall score cannot be adjudicated, and the note is the\nrationale the reference set depends on. Promote reviewed traces into a **versioned golden\ndataset** — the version is what lets validation state which labels a κ was computed against.\n\nWhere reviewers need the raw audio, image, or PDF to judge, run review over an **experiment**\nrather than a dataset, since experiments surface attachments more prominently; otherwise\nreviewers judge the pipeline's extraction instead of the system's behavior.\n\nWire two feeds: a stratified **production sample**, and the **scorer-disagreement queue** (items\nwhere scorers disagree or sit near a threshold) — the highest-value human attention available,\nat no extra collection cost. Reviewer identity, qualification, and timestamp go in item\nmetadata; fatigue drift and single-rater dominance are undetectable without them.\n",{"data":39,"body":40},{"name":4,"description":6},{"type":41,"children":42},"root",[43,52,75,82,102,108,170,176,199,205,233,239,268,272,298,310],{"type":44,"tag":45,"props":46,"children":48},"element","h1",{"id":47},"design-human-review-and-the-golden-set",[49],{"type":50,"value":51},"text","Design human review and the golden set",{"type":44,"tag":53,"props":54,"children":55},"p",{},[56,58,65,67,73],{"type":50,"value":57},"Contract: ",{"type":44,"tag":59,"props":60,"children":62},"code",{"className":61},[],[63],{"type":50,"value":64},"references\u002Finteraction-contract.md",{"type":50,"value":66},". Calibration, templates, provenance: ",{"type":44,"tag":59,"props":68,"children":70},{"className":69},[],[71],{"type":50,"value":72},"references\u002Freview-workflow.md",{"type":50,"value":74},".",{"type":44,"tag":76,"props":77,"children":79},"h2",{"id":78},"trigger",[80],{"type":50,"value":81},"Trigger",{"type":44,"tag":83,"props":84,"children":85},"ul",{},[86,92,97],{"type":44,"tag":87,"props":88,"children":89},"li",{},[90],{"type":50,"value":91},"\"Set up human review.\" \u002F \"Build a golden dataset.\" \u002F \"How should we adjudicate?\"",{"type":44,"tag":87,"props":93,"children":94},{},[95],{"type":50,"value":96},"A rubric that exists and now needs applying at scale by people.",{"type":44,"tag":87,"props":98,"children":99},{},[100],{"type":50,"value":101},"A judge that needs an anchor before it can be trusted.",{"type":44,"tag":76,"props":103,"children":105},{"id":104},"do",[106],{"type":50,"value":107},"Do",{"type":44,"tag":109,"props":110,"children":111},"ol",{},[112,125,130,141,153,165],{"type":44,"tag":87,"props":113,"children":114},{},[115,117,123],{"type":50,"value":116},"Identify the expertise required and the decision the labels will support. Draft the review\nform ",{"type":44,"tag":118,"props":119,"children":120},"strong",{},[121],{"type":50,"value":122},"before",{"type":50,"value":124}," asking about reviewer count.",{"type":44,"tag":87,"props":126,"children":127},{},[128],{"type":50,"value":129},"If no rubric or criteria exist yet, stop and elicit them from experts first. This skill\napplies criteria; it does not invent them.",{"type":44,"tag":87,"props":131,"children":132},{},[133,135,140],{"type":50,"value":134},"Select cases deliberately: representative items, plus the ambiguous region where systems\ndiffer, plus ",{"type":44,"tag":118,"props":136,"children":137},{},[138],{"type":50,"value":139},"every severe failure class",{"type":50,"value":74},{"type":44,"tag":87,"props":142,"children":143},{},[144,146,151],{"type":50,"value":145},"Collect a short ",{"type":44,"tag":118,"props":147,"children":148},{},[149],{"type":50,"value":150},"rationale and confidence",{"type":50,"value":152}," with every label — the rationales are where the\nrubric gets sharp.",{"type":44,"tag":87,"props":154,"children":155},{},[156,158,163],{"type":50,"value":157},"Have raters judge ",{"type":44,"tag":118,"props":159,"children":160},{},[161],{"type":50,"value":162},"independently before conferring",{"type":50,"value":164},", then adjudicate on the record. A panel\ndeferring to whoever speaks first discards the benefit of a panel.",{"type":44,"tag":87,"props":166,"children":167},{},[168],{"type":50,"value":169},"Report agreement and adjudicate before the set calibrates anything, then preserve it as a\nversioned dataset with per-item provenance and a refresh trigger.",{"type":44,"tag":76,"props":171,"children":173},{"id":172},"avoid",[174],{"type":50,"value":175},"Avoid",{"type":44,"tag":83,"props":177,"children":178},{},[179,184,189,194],{"type":44,"tag":87,"props":180,"children":181},{},[182],{"type":50,"value":183},"Do not treat majority opinion as ground truth without the relevant expertise.",{"type":44,"tag":87,"props":185,"children":186},{},[187],{"type":50,"value":188},"Do not force adjudication over genuine construct ambiguity — record it; it is a finding about\nthe rubric.",{"type":44,"tag":87,"props":190,"children":191},{},[192],{"type":50,"value":193},"Do not review only the easy middle; a golden set with no severe failures cannot validate a\nsafety scorer.",{"type":44,"tag":87,"props":195,"children":196},{},[197],{"type":50,"value":198},"Do not compare a scorer against these labels here.",{"type":44,"tag":76,"props":200,"children":202},{"id":201},"check",[203],{"type":50,"value":204},"Check",{"type":44,"tag":83,"props":206,"children":207},{},[208,213,218,223,228],{"type":44,"tag":87,"props":209,"children":210},{},[211],{"type":50,"value":212},"Reviewer qualifications stated and matched to the construct.",{"type":44,"tag":87,"props":214,"children":215},{},[216],{"type":50,"value":217},"Independent labeling before discussion; agreement computed, not assumed.",{"type":44,"tag":87,"props":219,"children":220},{},[221],{"type":50,"value":222},"Adjudication decisions recorded with reasons.",{"type":44,"tag":87,"props":224,"children":225},{},[226],{"type":50,"value":227},"Coverage spans clear successes, clear failures, ambiguous edges, severe failures.",{"type":44,"tag":87,"props":229,"children":230},{},[231],{"type":50,"value":232},"Provenance per item; set versioned; refresh trigger defined.",{"type":44,"tag":76,"props":234,"children":236},{"id":235},"risk",[237],{"type":50,"value":238},"Risk",{"type":44,"tag":83,"props":240,"children":241},{},[242,253,258,263],{"type":44,"tag":87,"props":243,"children":244},{},[245,247,252],{"type":50,"value":246},"Shared misunderstanding among reviewers produces a consistently wrong reference set that every\ndownstream number inherits — ",{"type":44,"tag":118,"props":248,"children":249},{},[250],{"type":50,"value":251},"high agreement is not validity",{"type":50,"value":74},{"type":44,"tag":87,"props":254,"children":255},{},[256],{"type":50,"value":257},"An unrepresentative review sample calibrates the scorer to the wrong distribution.",{"type":44,"tag":87,"props":259,"children":260},{},[261],{"type":50,"value":262},"Ceiling effects erase the benefit of multiple raters: if nearly every item is obvious,\nagreement is high and uninformative.",{"type":44,"tag":87,"props":264,"children":265},{},[266],{"type":50,"value":267},"Reviewer fatigue degrades labels across a long session, invisibly without timestamps.",{"type":44,"tag":76,"props":269,"children":270},{"id":8},[271],{"type":50,"value":9},{"type":44,"tag":53,"props":273,"children":274},{},[275,277,282,284,289,291,296],{"type":50,"value":276},"Run review in ",{"type":44,"tag":118,"props":278,"children":279},{},[280],{"type":50,"value":281},"human review queues",{"type":50,"value":283}," with ",{"type":44,"tag":118,"props":285,"children":286},{},[287],{"type":50,"value":288},"per-criterion scoring and free-text notes",{"type":50,"value":290},".\nPer-criterion is not optional: a single overall score cannot be adjudicated, and the note is the\nrationale the reference set depends on. Promote reviewed traces into a ",{"type":44,"tag":118,"props":292,"children":293},{},[294],{"type":50,"value":295},"versioned golden\ndataset",{"type":50,"value":297}," — the version is what lets validation state which labels a κ was computed against.",{"type":44,"tag":53,"props":299,"children":300},{},[301,303,308],{"type":50,"value":302},"Where reviewers need the raw audio, image, or PDF to judge, run review over an ",{"type":44,"tag":118,"props":304,"children":305},{},[306],{"type":50,"value":307},"experiment",{"type":50,"value":309},"\nrather than a dataset, since experiments surface attachments more prominently; otherwise\nreviewers judge the pipeline's extraction instead of the system's behavior.",{"type":44,"tag":53,"props":311,"children":312},{},[313,315,320,322,327],{"type":50,"value":314},"Wire two feeds: a stratified ",{"type":44,"tag":118,"props":316,"children":317},{},[318],{"type":50,"value":319},"production sample",{"type":50,"value":321},", and the ",{"type":44,"tag":118,"props":323,"children":324},{},[325],{"type":50,"value":326},"scorer-disagreement queue",{"type":50,"value":328}," (items\nwhere scorers disagree or sit near a threshold) — the highest-value human attention available,\nat no extra collection cost. Reviewer identity, qualification, and timestamp go in item\nmetadata; fatigue drift and single-rater dominance are undetectable without them.",{"items":330,"total":495},[331,347,363,377,389,405,419,431,445,461,473,481],{"slug":332,"name":332,"fn":333,"description":334,"org":335,"tags":336,"stars":344,"repoUrl":345,"updatedAt":346},"troubleshoot-braintrust-mcp","configure and troubleshoot Braintrust MCP servers","This plugin auto-configures a \"braintrust\" MCP server. If you can't see it or reach it, activate this skill\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[337,338,341],{"name":9,"slug":8,"type":16},{"name":339,"slug":340,"type":16},"Debugging","debugging",{"name":342,"slug":343,"type":16},"MCP","mcp",18,"https:\u002F\u002Fgithub.com\u002Fbraintrustdata\u002Fbraintrust-claude-plugin","2026-07-12T08:36:13.889274",{"slug":348,"name":348,"fn":349,"description":350,"org":351,"tags":352,"stars":27,"repoUrl":28,"updatedAt":362},"braintrust-analyze-eval-experiment","analyze LLM and agent eval experiments","Analyze completed LLM or agent eval experiments using uncertainty-aware and decision-relevant methods. Use to audit run completeness and pairing, calculate confidence intervals, run paired comparisons, report wins, losses, and ties, incorporate run-to-run variance, handle multiple comparisons, inspect subgroup performance, and test fragility to favorable slices. Use when results already exist and someone asks what they mean, whether a difference is real, or which model won. Do not use to design an experiment that has not yet collected results.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[353,356,357,358,359],{"name":354,"slug":355,"type":16},"Analysis","analysis",{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":360,"slug":361,"type":16},"Statistics","statistics","2026-08-20T03:53:01.13806",{"slug":364,"name":364,"fn":365,"description":366,"org":367,"tags":368,"stars":27,"repoUrl":28,"updatedAt":376},"braintrust-attribute-multi-variable-change","attribute performance changes to multiple variables","Attribute an observed change when several things moved at once — model plus prompt plus tools, a provider migration, a framework upgrade, or a vendor swap that bundles serving stack with model. Use when asked which part of a change caused the result, when a comparison's arms differ in more than one way, when a treatment has no uniform implementation across vendors, or when a serving-stack difference is confounded with a model difference. Do not use for a clean single-variable comparison, or to design an experiment that has not yet run.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[369,370,371,372,373],{"name":354,"slug":355,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":374,"slug":375,"type":16},"Performance","performance","2026-08-20T03:53:40.036077",{"slug":378,"name":378,"fn":379,"description":380,"org":381,"tags":382,"stars":27,"repoUrl":28,"updatedAt":388},"braintrust-build-eval-dataset","create and manage LLM eval datasets","Create, edit, audit, or compare eval datasets for LLM applications and agents, including target-population definition, case sourcing from production traces, stratified sampling, label provenance and label audits, expected values as constraints for open-ended tasks, dev\u002Ftest splits, contamination and leakage controls, headroom checks, refresh policy, and datasheets. Use when working on the content or lifecycle of an eval dataset. Do not use for sample-size or power calculations, scorer implementation, or open-ended adversarial failure discovery.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[383,384,385,386,387],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":14,"slug":15,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},"2026-08-20T03:53:33.304815",{"slug":390,"name":390,"fn":391,"description":392,"org":393,"tags":394,"stars":27,"repoUrl":28,"updatedAt":404},"braintrust-define-eval-objective","define LLM evaluation objectives","Create, edit, or audit an eval objective by working backward from a product decision to the target outcome, construct, population, intended claim, and verification-versus-validation questions. Use when a team is unsure what an eval should establish, asks \"what are we actually trying to measure,\" \"is this eval measuring the right thing,\" \"does this benchmark support our claim,\" or needs to turn a product goal into an eval objective and state which claims are out of scope. Do not use to select detailed metrics, design datasets, or implement scorers.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[395,396,397,398,401],{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":399,"slug":400,"type":16},"Product Management","product-management",{"name":402,"slug":403,"type":16},"Strategy","strategy","2026-08-20T03:53:00.07097",{"slug":406,"name":406,"fn":407,"description":408,"org":409,"tags":410,"stars":27,"repoUrl":28,"updatedAt":418},"braintrust-define-eval-release-gate","configure release gates for LLM applications","Create, edit, audit, or apply release gates for LLM applications and agents. Use to combine minimum meaningful improvement, statistical significance, regression rate, subgroup consistency, worst-run stability, all-attempts reliability, safety upper bounds, latency, and cost into an explicit ship-or-hold policy, to turn metrics into a CI gate, or to explain why a candidate failed a gate and what evidence would justify reconsideration. Do not use for general result analysis without a deployment decision.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[411,412,413,416,417],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":414,"slug":415,"type":16},"CI\u002FCD","ci-cd",{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},"2026-08-20T03:53:15.686158",{"slug":420,"name":420,"fn":421,"description":422,"org":423,"tags":424,"stars":27,"repoUrl":28,"updatedAt":430},"braintrust-deploy-evaluator","deploy evaluators to Braintrust","Take a validated scorer or classifier from definition to running instrument in Braintrust — scope selection, inline testing before saving, saving as an evaluator, attaching an online-scoring rule, activating it for new traffic, and backfilling history with a rewind. Use when a scorer needs to actually run against production logs, when an online-scoring rule needs to be created or changed, or when historical traces need scoring. Do not use to decide what the scorer should measure, to write its rubric, or to establish that it agrees with human judgment.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[425,426,429],{"name":9,"slug":8,"type":16},{"name":427,"slug":428,"type":16},"Deployment","deployment",{"name":21,"slug":22,"type":16},"2026-08-20T03:53:32.558937",{"slug":432,"name":432,"fn":433,"description":434,"org":435,"tags":436,"stars":27,"repoUrl":28,"updatedAt":444},"braintrust-design-eval-experiment","design controlled LLM eval experiments","Design or audit controlled eval experiments for model, prompt, retrieval, tool, guardrail, or agent-architecture changes. Use before data collection to state directional and minimum-effect hypotheses, name independent, dependent, and control variables including the serving environment and tool surface, choose paired designs, set repetitions and allocation, distinguish exploratory from confirmatory comparisons, and pre-specify stopping, exclusion, multiplicity, and analysis rules. Do not use primarily to analyze results already collected.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[437,438,439,442,443],{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":440,"slug":441,"type":16},"Experiments","experiments",{"name":18,"slug":19,"type":16},{"name":402,"slug":403,"type":16},"2026-08-20T03:53:36.534554",{"slug":446,"name":446,"fn":447,"description":448,"org":449,"tags":450,"stars":27,"repoUrl":28,"updatedAt":460},"braintrust-design-eval-instrumentation","design trace and evaluation dataset schemas","Design the trace and eval-dataset schema for an LLM app or agent, and wire the system to emit it. Use when deciding what to log, designing a trace schema, setting up tracing or observability before evals, or when failures cannot be debugged or sliced from existing traces — covering inputs, outputs, spans for tool and LLM calls, state changes, metadata, resolved configuration, serving path, tool manifest, per-item status, attachments, and subgroup variables. Do not use to decide what the evidence should mean, or to write scorers.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[451,452,453,454,457],{"name":9,"slug":8,"type":16},{"name":14,"slug":15,"type":16},{"name":21,"slug":22,"type":16},{"name":455,"slug":456,"type":16},"Observability","observability",{"name":458,"slug":459,"type":16},"Tracing","tracing","2026-08-20T03:53:37.274703",{"slug":462,"name":462,"fn":463,"description":464,"org":465,"tags":466,"stars":27,"repoUrl":28,"updatedAt":472},"braintrust-design-eval-metric-bundle","create multi-objective evaluation metric bundles","Create, edit, audit, or compare a multi-objective eval metric bundle covering product quality, safety, reliability, latency, and cost. Use to choose metrics for an eval, define a goodness bundle, distinguish optimization metrics from non-regression guardrails, expose tradeoffs, audit a KPI or single composite score for Goodhart and metric-gaming risk, or answer \"what should improve and what must not regress.\" Do not use to design trace schemas, build datasets, or implement scoring methods.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[467,468,469,470,471],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":374,"slug":375,"type":16},"2026-08-20T03:53:16.078609",{"slug":4,"name":4,"fn":5,"description":6,"org":474,"tags":475,"stars":27,"repoUrl":28,"updatedAt":29},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[476,477,478,479,480],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":14,"slug":15,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"slug":482,"name":482,"fn":483,"description":484,"org":485,"tags":486,"stars":27,"repoUrl":28,"updatedAt":494},"braintrust-discover-agent-failures","identify and classify agent failure modes","Search open-endedly for unanticipated agent failure modes and convert them into a named taxonomy and durable regression items. Use for requests to find out what goes wrong, surface unknown or silent failures, do error analysis over traces, cluster and triage production failures, or build a failure taxonomy — where the goal is discovering modes nobody thought to test rather than measuring a predefined criterion. Produces datasets and taxonomies, not headline scores. Do not use for adversarial attacks against a threat model, or to measure a known criterion.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[487,488,489,490,491],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":339,"slug":340,"type":16},{"name":21,"slug":22,"type":16},{"name":492,"slug":493,"type":16},"Triage","triage","2026-08-20T03:53:00.423941",27,{"items":497,"total":552},[498,506,514,522,530,538,544],{"slug":348,"name":348,"fn":349,"description":350,"org":499,"tags":500,"stars":27,"repoUrl":28,"updatedAt":362},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[501,502,503,504,505],{"name":354,"slug":355,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":360,"slug":361,"type":16},{"slug":364,"name":364,"fn":365,"description":366,"org":507,"tags":508,"stars":27,"repoUrl":28,"updatedAt":376},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[509,510,511,512,513],{"name":354,"slug":355,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":374,"slug":375,"type":16},{"slug":378,"name":378,"fn":379,"description":380,"org":515,"tags":516,"stars":27,"repoUrl":28,"updatedAt":388},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[517,518,519,520,521],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":14,"slug":15,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"slug":390,"name":390,"fn":391,"description":392,"org":523,"tags":524,"stars":27,"repoUrl":28,"updatedAt":404},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[525,526,527,528,529],{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":399,"slug":400,"type":16},{"name":402,"slug":403,"type":16},{"slug":406,"name":406,"fn":407,"description":408,"org":531,"tags":532,"stars":27,"repoUrl":28,"updatedAt":418},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[533,534,535,536,537],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":414,"slug":415,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"slug":420,"name":420,"fn":421,"description":422,"org":539,"tags":540,"stars":27,"repoUrl":28,"updatedAt":430},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[541,542,543],{"name":9,"slug":8,"type":16},{"name":427,"slug":428,"type":16},{"name":21,"slug":22,"type":16},{"slug":432,"name":432,"fn":433,"description":434,"org":545,"tags":546,"stars":27,"repoUrl":28,"updatedAt":444},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[547,548,549,550,551],{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":440,"slug":441,"type":16},{"name":18,"slug":19,"type":16},{"name":402,"slug":403,"type":16},24]