references/clinical_tables_tool_reference.md

Source: src/tooluniverse/data/clinical_tables_tools.json. All six tools are type ClinicalTablesTool, wrap NLM's Clinical Table Search Service, and require no API key. Live-tested (tu test + tu run with custom params); every field below was actually observed in a live response, not assumed from the JSON schema alone. Fields present live but absent from the JSON return_schema are flagged explicitly.

RxTerms_search_drugs

• Parameters: terms (string, required), max_results (int, default 20, max 500).
• tu test result: PASS ({"terms": "metformin", "max_results": 5}).
• Custom call: {"terms": "metfor", "max_results": 3}

  {
    "total_count": 20, "count": 3, "search_terms": "metfor",
    "results": [
      {"code": "metFORMIN (Oral Pill)", "DISPLAY_NAME": "metFORMIN (Oral Pill)",
       "STRENGTHS_AND_FORMS": ["  500 mg Tab", "  625 mg Tab", "  750 mg Tab", "  850 mg Tab", "1,000 mg Tab"],
       "RXCUIS": ["861007", "861021", "2703582", "861010", "861004"]},
      {"code": "metFORMIN XR (Oral Pill)", "...": "..."},
      {"code": "glyBURIDE/metFORMIN (Oral Pill)", "...": "..."}
    ]
  }
  

  Note STRENGTHS_AND_FORMS[i] corresponds positionally to RXCUIS[i] (same index = same strength/form's RxCUI) — confirmed by array length matching in every result. code and DISPLAY_NAME are identical in every observed result — code is not a separate opaque ID here, it's the display name used as the lookup key.

HealthConditions_search

• Parameters: terms (string, required), max_results (int, default 20, max 500).
• tu test result: PASS ({"terms": "diabetes", "max_results": 5}).
• Custom call: {"terms": "high blood pressure", "max_results": 3} — confirms informal-phrase -> formal-condition resolution works, not just prefix match on the formal name:

  {
    "total_count": 13, "count": 3, "search_terms": "high blood pressure",
    "results": [
      {"code": "374", "primary_name": "Hypertension",
       "consumer_name": "High blood pressure (hypertension (HTN))",
       "icd10cm_codes": "I10", "term_icd9_code": null},
      {"code": "3850", "primary_name": "Hypertension - essential",
       "consumer_name": "Hypertension - essential",
       "icd10cm_codes": "I10", "term_icd9_code": "401.9"},
      {"code": "11185", "primary_name": "Hypertension - benign essential",
       "consumer_name": "Hypertension - benign essential",
       "icd10cm_codes": "I10", "term_icd9_code": "401.1"}
    ]
  }
  

  term_icd9_code is frequently null (ICD-9 retired 2015; not every condition has a legacy mapping) — treat as optional, not always present.

DiseaseNames_search

• Parameters: terms (string, required), max_results (int, default 20, max 500).
• tu test result: PASS ({"terms": "cystic fibrosis", "max_results": 5}).
• Custom call: {"terms": "lupus", "max_results": 3}

  {
    "total_count": 62, "count": 3, "search_terms": "lupus",
    "results": [
      {"code": "C4551515", "primary_name": "Chilblain lupus"},
      {"code": "C4321325", "primary_name": "Lupus anticoagulant"},
      {"code": "C0409974", "primary_name": "Lupus erythematosus"}
    ]
  }
  

  code is the UMLS CUI (format C\d{7}). A broad term like "lupus" has 62 total matches spanning related-but-distinct concepts (an anticoagulant test, several lupus subtypes) — do not assume the first/top result is the one meant; disambiguate with the user if the term is ambiguous.

HCPCS_search

• Parameters: terms (string, required), max_results (int, default 20, max 500).
• tu test result: PASS ({"terms": "wheelchair", "max_results": 5}).
• Custom call: {"terms": "ambulance", "max_results": 3}

  {
    "total_count": 22, "count": 3, "search_terms": "ambulance",
    "results": [
      {"code": "A0424", "display": "Extra ambulance attendant"},
      {"code": "A0888", "display": "Noncovered ambulance mileage"},
      {"code": "A0998", "display": "Ambulance response/treatment"}
    ]
  }
  

  Matches on words anywhere in the description, not just a leading prefix (all three results have "ambulance" as a non-first word or standalone).

StarAlleles_search

• Parameters: terms (string, required), max_results (int, default 20, max 500).
• tu test result: PASS ({"terms": "CYP2D6", "max_results": 5}).
• Custom call: {"terms": "CYP2C19", "max_results": 3}

  {
    "total_count": 49, "count": 3, "search_terms": "CYP2C19",
    "results": [
      {"code": "CYP2C19*1A", "allele": "CYP2C19*1A", "unused": "None",
       "nucleotide_change": "None", "alternate_name": "", "protein_change": "None"},
      {"code": "CYP2C19*1B", "allele": "CYP2C19*1B", "unused": "99C>T; 991A>G",
       "nucleotide_change": "99C>T; 80161A>G", "alternate_name": "", "protein_change": "I331V"},
      {"code": "CYP2C19*1C", "allele": "CYP2C19*1C", "unused": "991A>G",
       "nucleotide_change": "80161A>G", "alternate_name": "", "protein_change": "I331V"}
    ]
  }
  

  Live-observed field not in the JSON return_schema: every result also includes an unused field (a secondary/legacy nucleotide-change notation that differs from nucleotide_change's values in the examples above — cause of the discrepancy not determined; prefer nucleotide_change as the documented field, but don't discard unused if the user's question hinges on matching a specific historical notation). "None"/"" are used interchangeably as "no value" across different fields — check for both when testing for presence/absence.
• 49 total matches for "CYP2C19" alone — this table is allele-definition granular (one row per named star allele), so a whole-gene query returns many rows; narrow with a specific allele ("CYP2C19*17") if only one definition is needed.

NPIProvider_search

• Parameters: terms (string, required), organization (boolean, nullable, default false), max_results (int, default 20, max 500).
• tu test result: PASS, 2 examples ({"terms": "Smith", "max_results": 5} and {"terms": "Mayo Clinic", "organization": true, "max_results": 5}).
• Custom call: {"terms": "Cleveland Clinic", "organization": true, "max_results": 3}

  {
    "total_count": 1216, "count": 3, "search_terms": "Cleveland Clinic",
    "results": [
      {"code": "1447487616", "name.full": "CLEVELAND CLINIC- CLEVELAND-OH", "NPI": "1447487616",
       "provider_type": "Hospital-General", "addr_practice.full": "9500 EUCLID AVENUE, CLEVELAND, OH 44195",
       "is_organization": true},
      {"code": "1548430275", "name.full": "CLEVELAND CLINIC", "NPI": "1548430275",
       "provider_type": "Hospital-General", "addr_practice.full": "A30 9500 EUCLID AVENUE, CLEVELAND, OH 44195",
       "is_organization": true},
      {"code": "1992990618", "name.full": "CLEVELAND CLINIC", "NPI": "1992990618",
       "provider_type": "Clinic or Group Practice", "addr_practice.full": "29800 BAINBRIDGE RD, SOLON, OH 44139",
       "is_organization": true}
    ]
  }
  

  code and NPI are identical in every observed result (redundant, both present). total_count of 1216 for "Cleveland Clinic" confirms this is a large, multi-site health system in the registry — a name search alone will not uniquely identify one facility; use addr_practice.full or the specific NPI to disambiguate a single location. Downstream link (from the tool's own JSON description, not independently verified here): the NPI value returned is what CMSOpenPayments_search_payments expects as its npi filter — use this tool to resolve a provider name to an NPI before querying payment history there.

---

ICD and RxNorm Tool Reference

Source: src/tooluniverse/data/icd_tools.json (5 tools, type ICDTool/ ICD10Tool) and src/tooluniverse/data/rxnorm_extended_tools.json (5 tools). Live-tested (tu test + tu run); every field below was actually observed live except where noted as schema-only.

ICD10_search_codes

• Parameters: query (string, required), limit (int, default 20, max 100).
• tu test result: PASS, 4/4 examples.
• Custom call: {"query": "type 2 diabetes", "limit": 3}

  {"data": {"total": 94, "results": [
    {"code": "E11.65", "name": "Type 2 diabetes mellitus with hyperglycemia"},
    {"code": "E11.9", "name": "Type 2 diabetes mellitus without complications"},
    {"code": "E10.A2", "name": "Type 1 diabetes mellitus, presymptomatic, Stage 2"}
  ]}, "metadata": {"source": "NLM Clinical Tables - ICD-10-CM",
    "version": "2026 ICD-10-CM codes"}}
  

  No API key. query also accepts a partial code (e.g. "E11"), per the tool's own test examples.

ICD10_get_code_info

• Parameters: code (string, required, e.g. "E11.9").
• tu test result: PASS, 3/3 examples.
• Custom call: {"code": "E11.9"} -> {"data": {"total": 1, "results": [{"code": "E11.9", "name": "Type 2 diabetes mellitus without complications"}]}}. One exact code in, one (or zero, if invalid) result out — this is a lookup, not a search.

ICD11_search_diseases / ICD11_get_entity / ICD11_browse_hierarchy

• Parameters (per schema): query/entity_id (required), plus linearization (mms/icf/ichi, default mms), flatResults, useFlexisearch, language.
• Live status: all 3 currently fail without setup. tu test ICD11_search_diseases -> 3/3 failed, every one returning {"status": "error", "error": "ICD API authentication required. Set ICD_CLIENT_ID and ICD_CLIENT_SECRET environment variables. Register at: https://icd.who.int/icdapi"}. Same for ICD11_get_entity (2/2 failed) and ICD11_browse_hierarchy (1/1 failed). This matches the tool's own declared required_api_keys, so it's expected behavior, not a bug — but note these 3 tools still appear in tu list/load normally (unlike, e.g., Addgene's tools elsewhere in this repo, which are excluded from tu list entirely until their key is set) because required_api_keys is declared inside this tool's parameter block rather than at the top level of the tool config — ToolUniverse's loader-level key gate only reads the top-level field, so it doesn't catch this case; the tool's own run() code still correctly refuses to call out without the keys. Net effect for a user is the same either way (a clear error, no fabricated data), just via a different code path than the more common gating pattern. Register free at https://icd.who.int/icdapi to use these.

RxNorm_find_rxcui

• Parameters: drug_name (string, required).
• tu test result: PASS, 3/3 examples.
• Custom call: {"drug_name": "metformin"} -> {"data": {"drug_name": "metformin", "rxcuis": ["6809"], "primary_rxcui": "6809", "found": true}}.

RxNorm_get_drug_info

• Parameters: rxcui (string) or drug_name (string) — at least one.
• tu test result: PASS, 3/3 examples.
• Custom call: {"rxcui": "6809"} -> {"data": {"rxcui": "6809", "name": "metformin", "synonym": null, "term_type": "IN", "term_type_label": "Ingredient (generic)", "language": "ENG"}}. term_type is an RxNorm TTY abbreviation — the response's own metadata.tty_key (see RxNorm_get_related_drugs below) maps every abbreviation to its full label, so no separate reference is needed.

RxNorm_get_related_drugs

• Parameters: rxcui and/or drug_name, optional tty (filter to one term-type, e.g. "BN" for brand name).
• tu test result: PASS, 3/3 examples.
• Custom call: {"rxcui": "6809", "tty": "BN"} -> 14 real branded products containing metformin (Janumet, Glucophage, Kombiglyze, Jentadueto, Kazano, Invokamet, Xigduo, Synjardy, Segluromet, Trijardy, Zituvimet, Riomet, Actoplus Met, Glumetza), each with its own rxcui. Response includes metadata.tty_key, a live-observed field not in the JSON schema — a full label map for every TTY code (IN=Ingredient generic, PIN=Precise Ingredient, BN=Brand Name, SCD=Semantic Clinical Drug, SBD=Semantic Branded Drug, GPCK/BPCK =Generic/Branded Pack, SCDF/SBDF=Clinical/Branded Drug Form, SCDC=Semantic Drug Component, MIN=Multiple Ingredients, DF=Dose Form).

RxNorm_get_ndc_status_history

• Parameters: ndc (string, required, 11-digit NDC).
• tu test result: PASS, 2/2 examples. Real observed shape (from tu test): {"ndc": "00093005801", "found": true, "ndc11": "00093005801", "status": "ACTIVE", "active": ...} — includes historical RxCUI remapping timeline per the tool's description; confirm exact remapping-field names against a live call before depending on them, only ndc/ndc11/status/active/found were directly inspected here.

RxNorm_get_ndc_properties

• Parameters: ndc (string, required — accepts common NDC formats, e.g. "0781-1506-10", not just the bare 11-digit form).
• tu test result: PASS, 1/1 example. Real observed shape: {"ndc": "0781-1506-10", "found": true, "products": [{"ndc_item": "00781150610", "ndc10": ...}]} — note the response normalizes the input format into ndc_item/ndc10 fields inside each product entry.

NCI Thesaurus (NCIt)

NCIThesaurus_search

• Parameters: term (string, required), page_size (int, optional, default 10, max 100).
• tu test result: PASS, 2/2 examples.
• Custom call: {"term": "immunotherapy", "page_size": 5} -> 2 real hits: {"code": "C15262", "name": "Immunotherapy"}, {"code": "C308", "name": "Immunotherapeutic Agent"}. Response wraps hits in data[] plus metadata: {source, total_results, query}.

NCIThesaurus_get_concept

• Parameters: code (string, required, e.g. "C3224").
• tu test result: PASS, 2/2 examples.
• Custom call: {"code": "C4872"} (Breast Carcinoma) -> full definition, an 18-entry synonyms[] array (each {"name", "type" (e.g. SY/PT), "source"} — source varies per synonym: caDSR, CPTAC, CTEP, CTRP, FDA, GDC, HemOnc, NCI-GLOSS, NCI, NICHD), and a properties[] array of {"type", "value"} pairs including UMLS_CUI (C0678222), Semantic_Type (Neoplastic Process), Neoplastic_Status (Malignant), and Maps_To entries — note UMLS_CUI lives inside properties, not as a top-level field.

NCIThesaurus_get_children

• Parameters: code (string, required).
• tu test result: PASS, 2/2 examples (C3262 Neoplasm -> 3 children incl. "Neoplasm by Morphology"; C3224 Melanoma -> children incl. "Amelanotic Melanoma").
• Returns data[] of {"code", "name", "leaf"} — immediate children only, one level down.

NCIThesaurus_get_parents

• Parameters: code (string, required).
• tu test result: PASS, 2/2 examples (C1647 Trastuzumab -> parent "Anti-HER2 Monoclonal Antibody"; C4872 Breast Carcinoma -> parents incl. "Carcinoma", "Malignant Neoplasm").
• Immediate parents only (leaf is typically null for a parent node, not a leaf-status marker on itself).

NCIThesaurus_get_concept_maps

• Parameters: code (string, required).
• tu test result: PASS, 2/2 examples.
• Custom call: {"code": "C4872"} -> {"code": "C4872", "name": "Breast Carcinoma", "maps": [{"type": "Has Synonym", "target_name": "Breast cancer", "target_code": "10006187", "target_term_type": "LLT", "target_terminology": "MedDRA", "target_terminology_version": "18.1"}, {"target_terminology": "GDC", "target_name": "Breast Cancer", "target_code": "relationship_primary_diagnosis", "target_term_type": "PT"}]}, plus metadata: {total_maps, target_terminologies}. This is the ONLY tool that returns cross-vocabulary codes — get_concept never includes them.

LOINC

Cross-cutting limitation, verified live across all 4 tools below: the underlying clinicaltables loinc_items/v3 index does not publish SYSTEM, SCALE_TYP, CLASS, METHOD_TYP, TIME_ASPCT, STATUS, or COMMON_TEST_RANK for ANY code — every response returns these as empty strings and separately lists them in a fields_unavailable[] array with a fields_unavailable_note explaining this is a data-source gap, not a per-code fact. Rely on LONG_COMMON_NAME/COMPONENT/SHORTNAME instead; if a user specifically needs specimen type (SYSTEM), say this index can't provide it rather than reporting an empty string as "no specimen."

LOINC_search_tests

• Parameters: terms (string, required), max_results (int, default 20, max 500), exclude_copyrighted (bool, default true).
• tu test result: PASS, 3/3 examples.
• Custom call: {"terms": "hemoglobin A1c", "max_results": 3} -> total_count: 14, 3 results incl. 4548-4 ("Hemoglobin A1c/Hemoglobin.total in Blood"), 112870-1 (a panel), 55454-3 (marked "Deprecated" in its own name — check LONG_COMMON_NAME for a "Deprecated" prefix before using a hit).

LOINC_get_code_details

• Parameters: loinc_code (string, required, hyphenated format e.g. "2093-3").
• tu test result: PASS, 3/3 examples.
• Custom call: {"loinc_code": "4548-4"} -> SHORTNAME: "HbA1c MFr Bld", PROPERTY: "MFr", COMPONENT: "Hemoglobin A1c/Hemoglobin.total", with TIME_ASPCT/SYSTEM/SCALE_TYP/CLASS/STATUS/COMMON_TEST_RANK all empty and listed in fields_unavailable.

LOINC_get_answer_list

• Parameters: loinc_code (string, required — accepts a specific code or a search term matching multiple codes).
• tu test result: PASS, 3/3 examples.
• Custom call: {"loinc_code": "883-9"} (ABO group) -> datatype: "CNE" (answer MUST come from the list) and AnswerLists: [{"AnswerListId": "LL2419-1", "AnswerListName": "ABO group", "answers": [{"AnswerStringID": "LA19710-5", "DisplayText": "Group A", "SequenceNo": 1}, ...Group B/O/AB]}]. Top-level answer_lists_found (here 1) distinguishes "matched a code with an answer list" from "matched but no list exists" (datatypes like REAL/ST/DT correctly return no list, and a note field appears only when answer_lists_found is 0).

LOINC_search_forms

• Parameters: terms (string, required — matched against instrument NAMES only, e.g. "PHQ-9"/"GAD-7"/"MMSE"/a panel name), max_results (int, default 20, max 200).
• tu test result: PASS, 3/3 examples (PHQ-9 -> 2 hits; "depression" -> 20 of 23 total; "pain scale" -> 5 hits).
• Returns whole instruments/panels only — a single question or lab test never appears here even if its wording matches; use LOINC_search_tests for individual tests/questions. A note field appears only when nothing matched.

UniProt Disease and Keyword Vocabularies (uniprot_ref_tools.json)

Source: src/tooluniverse/data/uniprot_ref_tools.json. All 4 tools type UniProtRefTool, no API key required.

UniProtRef_search_diseases / UniProtRef_get_disease

• Parameters: search_diseases takes query (string, required) + size (int, default 10, max 25); get_disease takes disease_id (string, required, format DI-XXXXX).
• tu test result: PASS, 2/2 examples each.
• Custom calls: {"query": "breast cancer"} -> DI-03803 ("Breast cancer, lobular", acronym LBC), DI-01559 ("familial breast-ovarian cancer 1"), DI-02603 ("familial breast-ovarian cancer 2") among results, each with reviewed_protein_count. {"disease_id": "DI-01559"} -> full record: name: "Breast-ovarian cancer, familial, 1", cross_references incl. {"database": "MIM", "id": "604370"} and {"database": "MeSH", "id": "D001943"}, plus separate reviewed_protein_count/unreviewed_protein_count (only get_disease exposes the unreviewed count; search_diseases only returns reviewed_protein_count).

UniProtRef_search_keywords / UniProtRef_get_keyword

• Parameters: search_keywords takes query (string, required) + size (int, default 10, max 25); get_keyword takes keyword_id (string, required, format KW-XXXX).
• tu test result: PASS, 2/2 examples each.
• Custom calls: {"query": "zinc finger"} -> KW-0863 ("Zinc-finger", category Domain, 13,427 reviewed proteins) — note this is a distinct keyword/ID from {"keyword_id": "KW-0862"} ("Zinc", category Ligand, 43,994 reviewed + 6,677,859 unreviewed proteins; parents: [{"id": "KW-9993", "name": "Ligand"}]; go_mappings: [] for this particular keyword, though the schema supports GO-term mappings on keywords that have them, e.g. KW-0053 Apoptosis -> GO:0006915).

UniProt Reference Proteomes

See skills/tooluniverse-sequence-retrieval/SKILL.md for UniProtRef_get_proteome/UniProtRef_search_proteomes (whole-organism proteome metadata) — kept in that skill rather than here since it's a sequence/genome-data-retrieval task, not a clinical-terminology normalization task, despite sharing the same source JSON file as the disease/keyword tools above.

BioPortal (bioportal_tools.json)

All 4 tools type BioPortalTool, no API key required. get_concept/ get_hierarchy need a full IRI (http://purl.obolibrary.org/obo/DOID_1909 style), not a bare code — get that IRI from a prior search_ontology_terms or annotate_text result's full_id/concept_full_id field.

• BioPortal_search_ontology_terms — query (required), ontologies (comma-separated acronyms, null = all 900+), page_size. tu test PASS, 2/2 examples. {"query": "melanoma"} -> DOID_1909 ("melanoma"); {"query": "apoptosis", "ontologies": "GO"} -> GO_0097194 ("execution phase of apoptosis").
• BioPortal_get_concept — ontology (required, e.g. "DOID", "GO"), concept_id (required, full IRI). tu test PASS, 2/2.
• BioPortal_annotate_text — text (required, <=~500 words), ontologies (optional restrict), longest_only (bool, default true). tu test PASS, 2/2. Custom call {"text": "Patient presents with hypertension and type 2 diabetes mellitus, prescribed metformin.", "ontologies": "DOID,RXNORM"} ->

  {"status": "success", "data": [
    {"matched_text": "HYPERTENSION", "from": 23, "to": 34, "match_type": "PREF",
     "concept_label": "hypertension", "concept_id": "DOID_10763",
     "concept_full_id": "http://purl.obolibrary.org/obo/DOID_10763"},
    {"matched_text": "TYPE 2 DIABETES MELLITUS", "from": 40, "to": 63, "match_type": "PREF",
     "concept_label": "type 2 diabetes mellitus", "concept_id": "DOID_9352"},
    {"matched_text": "METFORMIN", "from": 77, "to": 85, "match_type": "PREF",
     "concept_label": "metformin", "concept_id": "6809",
     "concept_full_id": "http://purl.bioontology.org/ontology/RXNORM/6809"}
  ], "metadata": {"source": "BioPortal Annotator (NCBO)", "total_annotations": 3, "text_length": 86}}
  

  The RXNORM concept_id 6809 matches RxNorm_find_rxcui's primary_rxcui for "metformin" documented earlier in this file — cross-validated.
• BioPortal_get_hierarchy — ontology, concept_id (full IRI, both required), direction (children/parents/ancestors, default children), page_size. tu test PASS, 2/2: DOID_9351 (diabetes mellitus) children include DOID_9744 ("type 1 diabetes mellitus"); HP_0001250 (Seizure) ancestors include HP_0012638 ("Abnormal nervous system physiology").

FHIR Terminology Service (fhir_terminology_tools.json)

Both tools type FHIRTerminologyTool, no API key required.

• FHIRTerminology_lookup_code — system (required, e.g. "snomed" or a full FHIR system URI), code (required). tu test PASS, 2/2: {"system": "snomed", "code": "22298006"} -> {"code": "22298006", "display": "Myocardial infarction", "system": "http://snomed.info/sct"}; {"system": "snomed", "code": "73211009"} -> "Diabetes mellitus".
• FHIRTerminology_expand_valueset — value_set_url (required), limit (int, 1-500, default 50). tu test PASS, 2/2. Custom call {"value_set_url": "http://snomed.info/sct?fhir_vs=isa/22298006", "limit": 5} -> real subtype list: 1755008 ("Old myocardial infarction"), 15990001 ("Acute myocardial infarction of posterolateral wall"), plus more, all sharing "system": "http://snomed.info/sct". Use the isa/<code> value-set URL pattern for "every descendant of this SNOMED concept" — no other tool in this skill does SNOMED subsumption.

NCI Drug Dictionary (nci_drugdict_tools.json)

Both tools type BaseRESTTool, no API key required.

• NCIDrugDict_search — query (required), matchType ("Begins"/"Contains", default "Begins"), size (default 10). tu test PASS, 3/3: {"query": "imatinib"} -> 1 result (code name "CGP 57148"); {"query": "pembrolizumab", "matchType": "Contains"} -> 8 results incl. USBrandName "Keytruda"; {"query": "tamoxifen"} -> 1 result.
• NCIDrugDict_get_drug — term_id (integer, required — get from a search result). tu test PASS, 3/3: term_id 37862 (imatinib mesylate) and 695789 (pembrolizumab, alias "Keytruda") both return full alias lists (CodeName/Synonym/USBrandName/Abbreviation types).

NCI EVS — Non-NCIt Terminologies (nci_evs_tools.json)

Both tools type NCIEVSTool, no API key required. Explicitly excludes NCIt (use NCIThesaurus_* for that) and MedDRA (licensed, not available).

• NCIEVS_search_terminology — terminology (required, e.g. "ctcae5", "icd9cm", "radlex", "ndfrt", "medrt", "canmed"), term (required), limit (default 20, max 100). tu test PASS, 2/2: {"terminology": "ctcae5", "term": "neutropenia"} -> C143481 ("Febrile neutropenia"); {"terminology": "radlex", "term": "fracture"} -> RID4650 ("fracture"). Custom call {"terminology": "icd9cm", "term": "diabetes"} -> real legacy codes: 250 ("Diabetes mellitus", leaf), 253.5 ("Diabetes insipidus", not leaf), 250.3 ("Diabetes with other coma", leaf) — confirms icd9cm works as a legacy-code fallback distinct from the current-standard ICD10_* tools above.
• NCIEVS_get_concept — terminology, code (both required). tu test PASS, 1/1: {"terminology": "ctcae5", "code": "C143481"} -> full detail incl. active: true.