Skip to Content
Unified docs shell with shared Classifyre tokens and acid-green highlight accents.
DetectorsCustom

Custom

Schema-driven detector documentation.

CUSTOMactiveP061 params19 examples
Detector Metadata
Capability catalog entry from all_detectors.json.

Categories

CLASSIFICATIONCOMPLIANCE

Supported Asset Types

TXTTABLEURLIMAGE

Recommended Model

mDeBERTa-v3 + SetFit + GLiNER + HuggingFace transformers

Notes

User-defined rules and pipelines tailored to specific business needs. Supports regex, GLiNER2, LLM, text classification, image classification, feature extraction, and object detection pipelines.

Parameters
Configuration parameters for the Custom detector. Shared from `CustomDetectorConfig`.
ParameterTypeRequiredDescriptionDefaultConstraints
custom_detector_keystringYesStable key used to identify one custom detector instance
namestringYesUser-facing name of custom detector
descriptionstringNo
methodenumNoExecution method for custom detector logic Allowed values: RULESET, CLASSIFIER, ENTITY, PIPELINE
languagesarrayNo["de","en"]
languages[]stringNo
rulesetobjectNono extra properties
ruleset.regex_rulesarrayNo[]
ruleset.regex_rules[]objectNono extra properties
ruleset.regex_rules[].idstringYesStable ID for this regex rule
ruleset.regex_rules[].namestringYesDisplay name for this regex rule
ruleset.regex_rules[].patternstringYesRegular expression pattern
ruleset.regex_rules[].flagsstringNoRegex flags (for example i, m, s)
ruleset.regex_rules[].severityenumNoSeverity level of finding Allowed values: critical, high, medium, low, info
ruleset.keyword_rulesarrayNo[]
ruleset.keyword_rules[]objectNono extra properties
ruleset.keyword_rules[].idstringYesStable ID for this keyword rule
ruleset.keyword_rules[].namestringYesDisplay name for this keyword rule
ruleset.keyword_rules[].keywordsarrayYesKeyword set to matchmin items 1
ruleset.keyword_rules[].keywords[]stringYes
ruleset.keyword_rules[].case_sensitivebooleanNoWhether keyword matching is case-sensitivefalse
ruleset.keyword_rules[].severityenumNoSeverity level of finding Allowed values: critical, high, medium, low, info
classifierobjectNono extra properties
classifier.labelsarrayNo[]
classifier.labels[]objectNono extra properties
classifier.labels[].idstringYes
classifier.labels[].namestringYes
classifier.labels[].descriptionstringNo
classifier.zero_shot_modelstringNoMoritzLaurer/mDeBERTa-v3-base-mnli-xnli
classifier.hypothesis_templatestringNoThis text contains {}.
classifier.training_examplesarrayNo[]
classifier.training_examples[]objectNono extra properties
classifier.training_examples[].textstringYes
classifier.training_examples[].labelstringYes
classifier.training_examples[].acceptedbooleanNotrue
classifier.training_examples[].sourcestringNoOrigin of this example (editor/feedback/import)editor
classifier.min_examples_per_labelintegerNo8min 1
classifier.setfit_modelstringNosentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
entityobjectNono extra properties
entity.entity_labelsarrayNo[]
entity.entity_labels[]stringNo
entity.entity_descriptionsobjectNoOptional GLiNER2 schema descriptions keyed by entity label{}
entity.modelstringNofastino/gliner2-base-v1
extractorobjectNoOptional structured extraction — runs when detector firesno extra properties
extractor.enabledbooleanNotrue
extractor.fieldsarrayYesmin items 1
extractor.fields[]objectYesOne output field in the extraction schemano extra properties
extractor.fields[].namestringYesOutput field name — becomes a key in extracted_data JSON
extractor.fields[].descriptionstringNoHuman-readable hint for what this field captures
extractor.fields[].typeenumNoAllowed values: string, number, boolean, list[string], list[number]string
extractor.fields[].entity_labelstringNoGLiNER2 schema label used for extraction (ENTITY and CLASSIFIER methods)
extractor.fields[].regex_patternstringNoRegex with one named capture group (?P<value>...) for RULESET method
extractor.fields[].regex_flagsstringNoRegex flags: i=case-insensitive, m=multiline, s=dotalli
extractor.fields[].aggregateenumNoHow to aggregate multiple matches Allowed values: first, last, list, join, countlist
extractor.fields[].join_separatorstringNo,
extractor.fields[].min_confidencenumberNoMinimum GLiNER confidence for this field0.4min 0, max 1
extractor.fields[].requiredbooleanNoIf true, skip saving extraction when this field is emptyfalse
extractor.gliner_modelstringNofastino/gliner2-base-v1
extractor.content_limitintegerNoChars of content to pass to extractor (classifier matched_content is only 320 chars)4000min 320, max 8192
pipeline_schemaobjectNo
max_findingsinteger | nullNoMaximum number of findings to return per assetnull