Australian geoscience data for agents: reports, drillholes, geochemistry, WA prospectivity and SQL.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Use when the user asks to compare two to eight candidate locations. For each point returns the prospectivity score and tier, tenure status, and the counts of nearby mines, drillholes and WAMEX reports over the point, with a mapUrl; rows are sorted by score, highest first.
2-8 candidate points
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http agenticgeomate https://geomate.nlp-tlp.org/mcp
Use when the user asks to compare two to eight candidate locations. For each point returns the prospectivity score and tier, tenure status, and the counts of nearby mines, drillholes and WAMEX reports over the point, with a mapUrl; rows are sorted by score, highest first.
| Argument | Type | Description |
|---|---|---|
| metal | string | |
| points* | array | 2-8 candidate points |
| radius_km | number |
{
"type": "object",
"required": [
"points"
],
"properties": {
"metal": {
"type": "string",
"default": "Au"
},
"points": {
"type": "array",
"items": {
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
}
}
},
"description": "2-8 candidate points"
},
"radius_km": {
"type": "number",
"default": 10
}
}
}Use to look at one data-library dataset (`dataset_id` from list_datasets) before querying or reusing it. Returns its name, description, visibility, citation, source link and licence (check the licence before reusing it), the projects it belongs to, the published version (number, state, source kind), and its layer with fields, types, source CRS, bounds and feature count. For a dataset the user owns it also returns the uploaded file's name, size and type and the import report. Pass the layer id to query_features.
| Argument | Type | Description |
|---|---|---|
| dataset_id* | string |
{
"type": "object",
"required": [
"dataset_id"
],
"properties": {
"dataset_id": {
"type": "string"
}
}
}Use to see what a raster dataset holds before sampling it (`raster_id` from list_datasets or list_geophysics_grids). Returns its size, bands, data type, nodata, resolution, WGS84 extent, per-band min / max / mean / p2 / p98, and its classes when it is classified.
| Argument | Type | Description |
|---|---|---|
| raster_id* | string |
{
"type": "object",
"required": [
"raster_id"
],
"properties": {
"raster_id": {
"type": "string"
}
}
}Use when the user wants to open or download a public document found with search_exploration_reports. Read-only: give the hit's corpus and doc_key; returns the document's title and series and temporary download links to its files (it changes nothing). 'wamex': the report's original PDF, appendix ZIPs and an OCR'd copy; 'sarig': the report PDF with its abstract, tenements, holder and year; 'ebookshop': the publication's PDF; 'hylogger': the report PDF with its hole, State and NVCL dataset id; 'geolit': the paper's PDF with its journal and year, except for papers licensed for text mining only, which have no file to share.
| Argument | Type | Description |
|---|---|---|
| corpus* | string | Which corpus the key belongs to. |
| doc_key* | string | The hit's doc_key: a WAMEX anumber ('A12345'), a GSWA product code ('N19EB'), a geolit sha256, or a hylogger '<dataset id>.sp' key, or a sarig key ('env02577'). |
{
"type": "object",
"required": [
"corpus",
"doc_key"
],
"properties": {
"corpus": {
"enum": [
"wamex",
"ebookshop",
"geolit",
"hylogger",
"sarig"
],
"type": "string",
"description": "Which corpus the key belongs to."
},
"doc_key": {
"type": "string",
"description": "The hit's doc_key: a WAMEX anumber ('A12345'), a GSWA product code ('N19EB'), a geolit sha256, or a hylogger '<dataset id>.sp' key, or a sarig key ('env02577')."
}
}
}Use when a signed-in user asks for drillhole data as a file: every hole within radius_km (at most 10) of a point, or one hole, as a CSV set (a zip of collars, survey, desurveyed path, assays and lithology, with a README) or as one OMF file for Leapfrog, Deswik, Micromine and other 3D packages. Positions are lon/lat and MGA; assays in ppm. Returns a download link (it expires), the file name, its size, how many holes it holds and how many were left out (holes are taken in order of how much data they record). The same intervals come as JSON, without sign-in, from get_drillhole_traces.
| Argument | Type | Description |
|---|---|---|
| lat* | number | |
| lon* | number | |
| hole | string | One hole's id (the 3D view's id, from get_drillhole_traces or find_drillhole_intercepts). |
| name | string | The file's name (default: the place and radius). |
| format | string | |
| radius_km | number | At most 10. |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"hole": {
"type": "string",
"description": "One hole's id (the 3D view's id, from get_drillhole_traces or find_drillhole_intercepts)."
},
"name": {
"type": "string",
"description": "The file's name (default: the place and radius)."
},
"format": {
"enum": [
"csv",
"omf"
],
"type": "string",
"default": "csv"
},
"radius_km": {
"type": "number",
"default": 2,
"description": "At most 10."
}
}
}Use to check on one of the user's own download packages or estimates (the id from estimate_download or prepare_download). Read-only. Returns its state, its progress (files and bytes done, time left) and, for each finished ZIP part, a temporary link to download it that works for one hour. States: estimating, estimated, awaiting_approval, queued, building, ready, partial (some files missing; see missing), failed, cancelled, rejected, expired.
| Argument | Type | Description |
|---|---|---|
| id* | string |
{
"type": "object",
"required": [
"id"
],
"properties": {
"id": {
"type": "string"
}
}
}Use when the user wants to download the files of public exploration reports and publications, to count the files and their size first. It saves the count as an estimate, under an id that prepare_download takes; nothing is packaged, sent, charged or deleted. Give `documents` [{corpus, key}] from a search, `anumbers` for WAMEX reports, a GeoThought `project_id` of the user's, or an `area` (a polygon, or a point and radius) to cover the reports there plus that area's public data layers as CSV/GeoJSON; up to 5,000 documents. Returns an estimate id, the number of files, the total size (and the PDFs' size alone), the documents with no files, and the number of ZIP parts (about 10 GB each). Tell the user the size before preparing anything. If `state` is still `estimating`, follow it with download_status.
| Argument | Type | Description |
|---|---|---|
| area | object | {polygon: [[lon, lat], ...]} or {point: [lon, lat], radius_km} (at most 50 km); at most 25,000 km². |
| name | string | What to call the package (the ZIP and email use it). |
| anumbers | array | WAMEX report numbers, e.g. ["A12345", "A6467"]. |
| documents | array | Documents as {corpus, key}, e.g. {corpus: "wamex", key: "A12345"} or {corpus: "sarig", key: "env01234"}: the corpus and key a search or document tool returned. |
| project_id | string | A GeoThought project the user can see. |
{
"type": "object",
"properties": {
"area": {
"type": "object",
"properties": {
"point": {
"type": "array",
"items": {
"type": "number"
}
},
"polygon": {
"type": "array",
"items": {
"type": "array",
"items": {
"type": "number"
}
}
},
"radius_km": {
"type": "number"
}
},
"description": "{polygon: [[lon, lat], ...]} or {point: [lon, lat], radius_km} (at most 50 km); at most 25,000 km²."
},
"name": {
"type": "string",
"description": "What to call the package (the ZIP and email use it)."
},
"anumbers": {
"type": "array",
"items": {
"type": "string"
},
"maxItems": 5000,
"description": "WAMEX report numbers, e.g. [\"A12345\", \"A6467\"]."
},
"documents": {
"type": "array",
"items": {
"type": "object",
"required": [
"corpus",
"key"
],
"properties": {
"key": {
"type": "string"
},
"corpus": {
"type": "string"
}
}
},
"maxItems": 5000,
"description": "Documents as {corpus, key}, e.g. {corpus: \"wamex\", key: \"A12345\"} or {corpus: \"sarig\", key: \"env01234\"}: the corpus and key a search or document tool returned."
},
"project_id": {
"type": "string",
"description": "A GeoThought project the user can see."
}
}
}Use when the user wants a whole data layer as a file (GeoJSON or CSV) for their own analysis. Exports one layer from list_export_layers, optionally clipped to a bbox and filtered, and returns a presigned download URL, the row count and a small preview. Large layers (for example drillholes, 3.46M rows) require a bbox. inspect_point and get_geochem summarise around a point; this tool returns the rows themselves.
| Argument | Type | Description |
|---|---|---|
| bbox | array | [west, south, east, north] lon/lat clip (required for huge layers). |
| layer* | string | One of the layers from list_export_layers. |
| limit | integer | Cap the rows (<= the layer's max). |
| format | string | |
| filters | object | Layer-specific filters from list_export_layers, e.g. {"metal":"Au"}, {"commodity":"gold"}, {"year_from":2000}. |
{
"type": "object",
"required": [
"layer"
],
"properties": {
"bbox": {
"type": "array",
"items": {
"type": "number"
},
"description": "[west, south, east, north] lon/lat clip (required for huge layers)."
},
"layer": {
"type": "string",
"description": "One of the layers from list_export_layers."
},
"limit": {
"type": "integer",
"description": "Cap the rows (<= the layer's max)."
},
"format": {
"enum": [
"geojson",
"csv"
],
"type": "string",
"default": "geojson"
},
"filters": {
"type": "object",
"description": "Layer-specific filters from list_export_layers, e.g. {\"metal\":\"Au\"}, {\"commodity\":\"gold\"}, {\"year_from\":2000}."
}
}
}Use when the user asks for drill core scanned by HyLogger (the national NVCL library, as CoreFlagging shows it): holes by name, operator or project (q), around a point (lon, lat, radius_km), in a bbox, and/or in a State. Returns each hole's key, name, State, operator, project, scanned depth range, instrument, whether it has logged geology, gold metres, lon/lat, distance, and a CoreFlagging link. get_core_scan_hole with the key gives one hole's details.
| Argument | Type | Description |
|---|---|---|
| q | string | |
| lat | number | |
| lon | number | |
| bbox | array | |
| limit | integer | |
| state | string | WA, SA, NT, QLD, NSW, VIC or TAS. |
| radius_km | number | Around lon/lat; at most 200 km. |
{
"type": "object",
"properties": {
"q": {
"type": "string"
},
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"bbox": {
"type": "array",
"items": {
"type": "number"
},
"maxItems": 4,
"minItems": 4
},
"limit": {
"type": "integer",
"default": 25,
"maximum": 200
},
"state": {
"type": "string",
"description": "WA, SA, NT, QLD, NSW, VIC or TAS."
},
"radius_km": {
"type": "number",
"default": 10,
"description": "Around lon/lat; at most 200 km."
}
}
}Use when the user asks for downhole intercepts ("8 m @ 2.1 g/t Au from 120 m") of an element in the drillholes around a point. They are computed from the company assays the platform holds, by one stated rule: a run starts and ends on a sample at or above the cut-off, holds at most max_dilution_m metres in a row below it (unsampled gaps count as zero), keeps a length-weighted grade at or above it, and is at least min_width_m long. The default cut-offs are 0.5 g/t Au, 5 g/t Ag, 0.1 % Cu, 0.5 % Ni, 0.5 % Zn, 0.3 % Li2O, .... Returns the rule, how many holes were looked at and assayed for the element, the intercepts sorted by grade × width, largest first, with each hole's name, type, year, operator, WAMEX report and collar, and a link to the 3D view coloured by the element. Computed from reported assays; not a resource estimate.
| Argument | Type | Description |
|---|---|---|
| lat* | number | |
| lon* | number | |
| hole | string | Only this hole (its name, e.g. "MFED001"). |
| limit | integer | |
| element | string | |
| per_hole | integer | Intercepts kept per hole. |
| radius_km | number | Around lon/lat; at most 5 km. |
| cutoff_ppm | number | The cut-off in ppm (1 ppm = 1 g/t; 1 % = 10,000 ppm). |
| min_width_m | number | |
| max_dilution_m | number |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"hole": {
"type": "string",
"description": "Only this hole (its name, e.g. \"MFED001\")."
},
"limit": {
"type": "integer",
"default": 20,
"maximum": 100
},
"element": {
"enum": [
"Ag",
"As",
"Au",
"Ba",
"Be",
"Bi",
"Co",
"Cr",
"Cs",
"Cu",
"Fe",
"Li",
"Mn",
"Mo",
"Nb",
"Ni",
"P",
"Pb",
"Sb",
"Sn",
"Sr",
"Th",
"U",
"W",
"Y",
"Zn"
],
"type": "string",
"default": "Au"
},
"per_hole": {
"type": "integer",
"default": 3,
"maximum": 10,
"description": "Intercepts kept per hole."
},
"radius_km": {
"type": "number",
"default": 1,
"description": "Around lon/lat; at most 5 km."
},
"cutoff_ppm": {
"type": "number",
"description": "The cut-off in ppm (1 ppm = 1 g/t; 1 % = 10,000 ppm)."
},
"min_width_m": {
"type": "number",
"default": 1
},
"max_dilution_m": {
"type": "number",
"default": 2
}
}
}Use when the user asks which drillholes there are in a place or of a company or project (WA open-file and company collars). Filter by a bbox or a point and radius, and/or a company or project key from search_tenement_holders; drilling type groups (dd = diamond core, rc = RC, ac = aircore and RAB, shallow = vacuum, auger, costeans, percussion..., other); a minimum depth; drilled years. Returns the count, counts by hole type and by decade, the holes sorted by depth, deepest first (holeid, type, depth, year, operator, project, lon/lat), and a map link showing them. Assays and traces come from get_drillhole_traces or sql_query.
| Argument | Type | Description |
|---|---|---|
| lat | number | |
| lon | number | |
| bbox | array | [west, south, east, north] in degrees; at most 25 square degrees. |
| limit | integer | |
| types | array | |
| company | string | A company key from search_tenement_holders, e.g. "NORTHERN STAR". |
| project | string | A project key from search_tenement_holders, e.g. "jundee". |
| year_to | integer | |
| radius_km | number | Around lon/lat; at most 50 km. |
| year_from | integer | |
| min_depth_m | number |
{
"type": "object",
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"bbox": {
"type": "array",
"items": {
"type": "number"
},
"maxItems": 4,
"minItems": 4,
"description": "[west, south, east, north] in degrees; at most 25 square degrees."
},
"limit": {
"type": "integer",
"default": 50,
"maximum": 500
},
"types": {
"type": "array",
"items": {
"enum": [
"dd",
"rc",
"ac",
"shallow",
"other"
],
"type": "string"
}
},
"company": {
"type": "string",
"description": "A company key from search_tenement_holders, e.g. \"NORTHERN STAR\"."
},
"project": {
"type": "string",
"description": "A project key from search_tenement_holders, e.g. \"jundee\"."
},
"year_to": {
"type": "integer"
},
"radius_km": {
"type": "number",
"description": "Around lon/lat; at most 50 km."
},
"year_from": {
"type": "integer"
},
"min_depth_m": {
"type": "number"
}
}
}Use to locate an identifier or a place on the map: a tenement ("E 53/1209"), a WAMEX report ("A12345"), a drillhole id, a map sheet, a place name, or coordinates in decimal degrees ("-30.75, 121.47"), degrees-minutes-seconds ("30°45'S 121°28'12\"E") or MGA ("MGA51 352000 6600000"). Returns each match's kind, label, detail, lon/lat and bbox, with a map link that opens the point's summary card. Companies and projects are found with search_tenement_holders.
| Argument | Type | Description |
|---|---|---|
| q* | string | An identifier, place name or coordinates. |
| limit | integer |
{
"type": "object",
"required": [
"q"
],
"properties": {
"q": {
"type": "string",
"description": "An identifier, place name or coordinates."
},
"limit": {
"type": "integer",
"default": 10,
"maximum": 30
}
}
}Use to add documents to one of the user's projects: `documents` [{corpus, key, title?}] and/or `uploads` (dataset ids of their own files, from geothought_uploads or upload_data). Only the user's own files can be added; GeoThought reads at most 50 of one user's own files (2,000 pages) across their projects. Returns how many were added and the project's document count.
| Argument | Type | Description |
|---|---|---|
| project* | string | |
| uploads | array | |
| documents | array |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
},
"uploads": {
"type": "array",
"items": {
"type": "string"
}
},
"documents": {
"type": "array",
"items": {
"type": "object"
}
}
}
}Use to read a project's answers: for a page of its documents (offset, limit) or one `document` {corpus, key}. Returns each question's answer with the quote, page and box it came from. Optional `schema` (name) for its questions only.
| Argument | Type | Description |
|---|---|---|
| limit | integer | |
| offset | integer | |
| schema | string | |
| project* | string | |
| document | object |
{
"type": "object",
"required": [
"project"
],
"properties": {
"limit": {
"type": "integer",
"maximum": 200,
"minimum": 1
},
"offset": {
"type": "integer",
"minimum": 0
},
"schema": {
"type": "string"
},
"project": {
"type": "string"
},
"document": {
"type": "object"
}
}
}Use to turn a set of questions into a schema of a kind: GeoThought's model groups and orders them; nothing is saved. Give `question_ids` of saved questions and/or `questions` objects, and optionally a `goal`. Returns the arranged schema. A schema is {kind: mindmap|relation|thoughtgraph, name, description?, questions: [Question], tree?, edges?}. A Question is {id (short, unique in the schema), text, answer_type: boolean|enum|numeric|date|text|place|list|relation, label?, enum_values? (enum), unit? (numeric), relation? {subject_type, predicate, object_type} (relation), context?, temporal?, spatial?}. A mindmap's tree is [{title, question?: <question id>, children?: [...]}]; a thought graph's edges are [{source: <question id>, target: <question id>, condition}] (the answer that leads on).
| Argument | Type | Description |
|---|---|---|
| goal | string | |
| kind* | string | |
| questions | array | |
| question_ids | array |
{
"type": "object",
"required": [
"kind"
],
"properties": {
"goal": {
"type": "string"
},
"kind": {
"enum": [
"mindmap",
"relation",
"thoughtgraph"
],
"type": "string"
},
"questions": {
"type": "array",
"items": {
"type": "object"
}
},
"question_ids": {
"type": "array",
"items": {
"type": "string"
}
}
}
}Use when the user asks a question about a project's documents in words. GeoThought answers from the project's documents and saved answers. Returns the answer with each claim cited to a document page. For a follow-up, `earlier_questions` may hold up to 3 questions asked with this tool about the same project and the answers it gave; send nothing else from the conversation.
| Argument | Type | Description |
|---|---|---|
| project* | string | |
| question* | string | |
| earlier_questions | array |
{
"type": "object",
"required": [
"project",
"question"
],
"properties": {
"project": {
"type": "string"
},
"question": {
"type": "string"
},
"earlier_questions": {
"type": "array",
"items": {
"type": "object",
"required": [
"question",
"answer"
],
"properties": {
"answer": {
"type": "string"
},
"question": {
"type": "string"
}
}
},
"maxItems": 3
}
}
}Use to check a schema body before saving it. Returns what is wrong or missing in it (unreachable questions, bad edges, ...). A schema is {kind: mindmap|relation|thoughtgraph, name, description?, questions: [Question], tree?, edges?}. A Question is {id (short, unique in the schema), text, answer_type: boolean|enum|numeric|date|text|place|list|relation, label?, enum_values? (enum), unit? (numeric), relation? {subject_type, predicate, object_type} (relation), context?, temporal?, spatial?}. A mindmap's tree is [{title, question?: <question id>, children?: [...]}]; a thought graph's edges are [{source: <question id>, target: <question id>, condition}] (the answer that leads on).
| Argument | Type | Description |
|---|---|---|
| schema* | object |
{
"type": "object",
"required": [
"schema"
],
"properties": {
"schema": {
"type": "object"
}
}
}Use to turn a schema into another kind (mindmap, relation, thoughtgraph); nothing is saved. Returns the converted schema.
| Argument | Type | Description |
|---|---|---|
| kind* | string | |
| schema* | object |
{
"type": "object",
"required": [
"schema",
"kind"
],
"properties": {
"kind": {
"enum": [
"mindmap",
"relation",
"thoughtgraph"
],
"type": "string"
},
"schema": {
"type": "object"
}
}
}Use to copy a project (the user's or a public one) into a new private project of theirs; uploaded files of another account are left out. Returns the copy's id.
| Argument | Type | Description |
|---|---|---|
| project* | string |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
}
}
}Use when the user asks for a new GeoThought project. Creates a private project with a name, the schemas it reads with (refs from geothought_schemas / geothought_templates), optionally its documents ([{corpus, key}] from geothought_find_documents; the user's own files as {corpus: "upload", key: <dataset id>} or in `uploads`), corpora to search in, and a period. Returns its id.
| Argument | Type | Description |
|---|---|---|
| name* | string | |
| corpora | array | |
| schemas | array | |
| uploads | array | |
| year_to | integer | |
| documents | array | |
| year_from | integer | |
| description | string |
{
"type": "object",
"required": [
"name"
],
"properties": {
"name": {
"type": "string"
},
"corpora": {
"type": "array",
"items": {
"type": "string"
}
},
"schemas": {
"type": "array",
"items": {
"type": "string"
}
},
"uploads": {
"type": "array",
"items": {
"type": "string"
}
},
"year_to": {
"type": "integer"
},
"documents": {
"type": "array",
"items": {
"type": "object"
}
},
"year_from": {
"type": "integer"
},
"description": {
"type": "string"
}
}
}Use when the user asks to delete one of their projects, with its document list and its readings. Cannot be undone. Returns the deleted id.
| Argument | Type | Description |
|---|---|---|
| project* | string |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
}
}
}Use when the user asks to delete a saved question from their library. Returns the deleted id.
| Argument | Type | Description |
|---|---|---|
| id* | string |
{
"type": "object",
"required": [
"id"
],
"properties": {
"id": {
"type": "string"
}
}
}Use when the user asks to delete one of their schemas. Projects reading with it keep their answers. Returns the deleted id.
| Argument | Type | Description |
|---|---|---|
| id* | string |
{
"type": "object",
"required": [
"id"
],
"properties": {
"id": {
"type": "string"
}
}
}Use when the user asks to delete one of their files: it is taken out of every project, its answers deleted, and the file moved to their data library's trash. Returns what was deleted.
| Argument | Type | Description |
|---|---|---|
| key* | string |
{
"type": "object",
"required": [
"key"
],
"properties": {
"key": {
"type": "string"
}
}
}Use to fetch another page of sections of the document taken with geothought_next_document (page 0 is the first). Returns that page's numbered sections and their text.
| Argument | Type | Description |
|---|---|---|
| page* | integer | |
| task* | string | The task id geothought_next_document gave. |
| token* | string | The lease token that came with it. |
{
"type": "object",
"required": [
"task",
"token",
"page"
],
"properties": {
"page": {
"type": "integer",
"minimum": 0
},
"task": {
"type": "string",
"description": "The task id geothought_next_document gave."
},
"token": {
"type": "string",
"description": "The lease token that came with it."
}
}
}Use when the user describes what they want to find out in documents and needs a schema for it. GeoThought's model drafts one; nothing is saved. kind: mindmap (questions grouped as a tree), relation (relations between things) or thoughtgraph (a decision graph). With reuse_my_questions, the user's saved questions are reused where they fit. Returns the draft; review it with the user, then geothought_save_schema. A schema is {kind: mindmap|relation|thoughtgraph, name, description?, questions: [Question], tree?, edges?}. A Question is {id (short, unique in the schema), text, answer_type: boolean|enum|numeric|date|text|place|list|relation, label?, enum_values? (enum), unit? (numeric), relation? {subject_type, predicate, object_type} (relation), context?, temporal?, spatial?}. A mindmap's tree is [{title, question?: <question id>, children?: [...]}]; a thought graph's edges are [{source: <question id>, target: <question id>, condition}] (the answer that leads on).
| Argument | Type | Description |
|---|---|---|
| kind* | string | |
| description* | string | |
| reuse_my_questions | boolean |
{
"type": "object",
"required": [
"kind",
"description"
],
"properties": {
"kind": {
"enum": [
"mindmap",
"relation",
"thoughtgraph"
],
"type": "string"
},
"description": {
"type": "string"
},
"reuse_my_questions": {
"type": "boolean",
"default": true
}
}
}Use when the user asks to change the details of one of their own uploaded documents: its title, year, operator or commodity. Only the fields given are changed, and the new details show in every project that holds the document. Returns the document's key, title, processing state and the projects it is in.
| Argument | Type | Description |
|---|---|---|
| key* | string | |
| year | integer | null | |
| title | string | |
| operator | string | |
| commodity | string |
{
"type": "object",
"required": [
"key"
],
"properties": {
"key": {
"type": "string"
},
"year": {
"type": [
"integer",
"null"
]
},
"title": {
"type": "string"
},
"operator": {
"type": "string"
},
"commodity": {
"type": "string"
}
}
}Use to find documents to add to a project. mode=search: by words (`query`); mode=suggest: documents likely to answer the questions of `schemas` (refs) or of `project`'s schemas; mode=browse: a corpus's catalogue by words (`query`), `tags` (commodities, series), `who` (operator, author). Narrow by year_from/year_to. corpora default wamex. Returns documents with the corpus and key geothought_add_documents takes.
| Argument | Type | Description |
|---|---|---|
| who | string | |
| mode | string | |
| tags | array | |
| limit | integer | |
| query | string | |
| corpora | array | |
| project | string | |
| schemas | array | |
| year_to | integer | |
| year_from | integer |
{
"type": "object",
"properties": {
"who": {
"type": "string"
},
"mode": {
"enum": [
"search",
"suggest",
"browse"
],
"type": "string",
"default": "search"
},
"tags": {
"type": "array",
"items": {
"type": "string"
}
},
"limit": {
"type": "integer",
"default": 30,
"maximum": 200,
"minimum": 1
},
"query": {
"type": "string"
},
"corpora": {
"type": "array",
"items": {
"type": "string"
}
},
"project": {
"type": "string"
},
"schemas": {
"type": "array",
"items": {
"type": "string"
}
},
"year_to": {
"type": "integer"
},
"year_from": {
"type": "integer"
}
}
}Use to read a schema's body by its ref (s:<id> for the user's own, t:<id> for a template). Returns the schema.
| Argument | Type | Description |
|---|---|---|
| ref* | string |
{
"type": "object",
"required": [
"ref"
],
"properties": {
"ref": {
"type": "string"
}
}
}Use to summarise a project's answers to one schema. Returns, per question, the counts and values; relations; places and years; a thought graph's routes.
| Argument | Type | Description |
|---|---|---|
| schema* | string | |
| project* | string |
{
"type": "object",
"required": [
"project",
"schema"
],
"properties": {
"schema": {
"type": "string"
},
"project": {
"type": "string"
}
}
}Use to see which models the user can read and ask with (GeoThought's own and any they added on the website). Returns each model's id and name, and which ones the user chose.
No arguments.
{
"type": "object",
"properties": {}
}Use during an assistant reading (after geothought_start_agent_reading) to take the next document. Returns `task` and `token` (needed for every call about this document), the `instructions` and the `checklist` of questions (Q1..Qn), and the first `page` of its `sections` (numbered; `pages` says how many pages there are; geothought_document_sections fetches the others). Answer each section only from its text, with verbatim quotes, and send the replies with geothought_submit_replies. The document stays assigned for an hour after each call about it. Returns document: null when none is waiting.
| Argument | Type | Description |
|---|---|---|
| project* | string |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
}
}
}Use to look at one GeoThought project (the user's own or a public one). Returns its schemas, period, documents (each with its corpus and key; an uploaded file with how far it has been read: text, layout, OCR), whether it is the user's and whether it is public, and its share link if it has one.
| Argument | Type | Description |
|---|---|---|
| project* | string |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
}
}
}Use when the user asks which GeoThought projects they have. Returns each of the signed-in user's projects: id, name, how many documents, and the schemas (with their question counts) it reads with. Pass a project id to geothought_project or geothought_start_agent_reading.
No arguments.
{
"type": "object",
"properties": {}
}Use only when the user asks to make one of their projects public, or private again. public=true: anyone can then open the project, its documents (including the user's own uploaded files in it) and its answers, through the returned link and on GeoThought's public pages; allow_ask lets visitors ask it questions. public=false: private again, and the link stops working. Returns the link and a `tell_user` text that says what was opened or closed; give it to the user.
| Argument | Type | Description |
|---|---|---|
| public* | boolean | |
| project* | string | |
| allow_ask | boolean |
{
"type": "object",
"required": [
"project",
"public"
],
"properties": {
"public": {
"type": "boolean"
},
"project": {
"type": "string"
},
"allow_ask": {
"type": "boolean"
}
}
}Use to list the user's saved GeoThought questions (their question library). Returns each question's id, text and answer type.
No arguments.
{
"type": "object",
"properties": {}
}Use before reading a project to see what it would take. Returns the documents, questions, answers already there and what is left. Optional `schemas` (names) and `documents` [{corpus, key}].
| Argument | Type | Description |
|---|---|---|
| project* | string | |
| schemas | array | |
| documents | array |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
},
"schemas": {
"type": "array",
"items": {
"type": "string"
}
},
"documents": {
"type": "array",
"items": {
"type": "object"
}
}
}
}Use to check a project's reading. Returns its latest reading's state: documents done, being read and waiting, the reader, and a short reason if it failed.
| Argument | Type | Description |
|---|---|---|
| project* | string |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
}
}
}Use to hand a document back unread, e.g. when it is too long to read now; it is offered again later. Returns whether it was released.
| Argument | Type | Description |
|---|---|---|
| task* | string | The task id geothought_next_document gave. |
| token* | string | The lease token that came with it. |
| reason | string |
{
"type": "object",
"required": [
"task",
"token"
],
"properties": {
"task": {
"type": "string",
"description": "The task id geothought_next_document gave."
},
"token": {
"type": "string",
"description": "The lease token that came with it."
},
"reason": {
"type": "string"
}
}
}Use to take documents out of a project (`documents` [{corpus, key}], `uploads` ids, or all=true). The files themselves stay. Returns how many were removed and the project's document count.
| Argument | Type | Description |
|---|---|---|
| all | boolean | |
| project* | string | |
| uploads | array | |
| documents | array |
{
"type": "object",
"required": [
"project"
],
"properties": {
"all": {
"type": "boolean"
},
"project": {
"type": "string"
},
"uploads": {
"type": "array",
"items": {
"type": "string"
}
},
"documents": {
"type": "array",
"items": {
"type": "object"
}
}
}
}Use to save a question to the user's library; with `id`, it replaces that saved question. `question` is {text, answer_type: boolean|enum|numeric|date|text|place|list|relation, enum_values?, unit?, relation?, context?, temporal?, spatial?}. Schemas use saved questions (geothought_arrange_schema, geothought_draft_schema). Returns the saved question.
| Argument | Type | Description |
|---|---|---|
| id | string | |
| question* | object |
{
"type": "object",
"required": [
"question"
],
"properties": {
"id": {
"type": "string"
},
"question": {
"type": "object"
}
}
}Use to save a schema as the user's own (private); with `id`, it replaces that schema of theirs (the projects reading with it follow). Returns its ref (s:<id>) to put in a project. A schema is {kind: mindmap|relation|thoughtgraph, name, description?, questions: [Question], tree?, edges?}. A Question is {id (short, unique in the schema), text, answer_type: boolean|enum|numeric|date|text|place|list|relation, label?, enum_values? (enum), unit? (numeric), relation? {subject_type, predicate, object_type} (relation), context?, temporal?, spatial?}. A mindmap's tree is [{title, question?: <question id>, children?: [...]}]; a thought graph's edges are [{source: <question id>, target: <question id>, condition}] (the answer that leads on).
| Argument | Type | Description |
|---|---|---|
| id | string | |
| name | string | |
| schema* | object |
{
"type": "object",
"required": [
"schema"
],
"properties": {
"id": {
"type": "string"
},
"name": {
"type": "string"
},
"schema": {
"type": "object"
}
}
}Use to list the user's own schemas (refs s:<id>). Returns each one's ref, name, kind, visibility and question count.
No arguments.
{
"type": "object",
"properties": {}
}Use when the user wants this assistant, rather than GeoThought's own model, to read a project's documents and answer its schemas' questions. Queues the documents that have no answer yet; the answers given are saved with the project. Optional `schemas` (names) to ask only some of its schemas' questions, `documents` (report keys such as A24950) to read only some, `limit` to queue only that many documents. Returns the reading's id and size; continue with geothought_next_document. Reading costs the assistant model's tokens: roughly 10k-30k input tokens a document for a schema of a few dozen questions.
| Argument | Type | Description |
|---|---|---|
| limit | integer | |
| project* | string | |
| schemas | array | |
| documents | array |
{
"type": "object",
"required": [
"project"
],
"properties": {
"limit": {
"type": "integer",
"maximum": 300,
"minimum": 1
},
"project": {
"type": "string"
},
"schemas": {
"type": "array",
"items": {
"type": "string"
}
},
"documents": {
"type": "array",
"items": {
"type": "string"
}
}
}
}Use when the user asks GeoThought's own reader to read one of their projects: every document is asked its schemas' questions and the answers are kept with their page and box. `model`: a model id from geothought_models (default: the user's choice, else GeoThought's own). Optional `schemas` (names), `documents` [{corpus, key}], `limit`, `force` (read again what was read, replacing those answers). A model the user added on the website may be an outside service: the documents' text is sent to it. Returns the reading's state; follow it with geothought_reading_status.
| Argument | Type | Description |
|---|---|---|
| force | boolean | |
| limit | integer | |
| model | string | |
| project* | string | |
| schemas | array | |
| documents | array |
{
"type": "object",
"required": [
"project"
],
"properties": {
"force": {
"type": "boolean"
},
"limit": {
"type": "integer",
"minimum": 1
},
"model": {
"type": "string"
},
"project": {
"type": "string"
},
"schemas": {
"type": "array",
"items": {
"type": "string"
}
},
"documents": {
"type": "array",
"items": {
"type": "object"
}
}
}
}Use when the user asks to stop a project's reading. Documents not yet taken are dropped; answers already saved stay. Returns the reading's final state.
| Argument | Type | Description |
|---|---|---|
| project* | string |
{
"type": "object",
"required": [
"project"
],
"properties": {
"project": {
"type": "string"
}
}
}Use to send the answers for some or all sections of the document being read: `replies` is {"<section number>": {"Q1": {"status": "found", "value": ..., "quote": "verbatim from that section"}, ...}}, leaving out the questions a section does not answer, and {} for a section with nothing (so it counts as read). Replies can be sent a page at a time; a section sent again replaces its reply. With done=true the replies are checked and saved as the document's answers (an answer whose quote is not in its section is dropped; the result says how many). `model`: the model the assistant runs on, recorded with the answers. Returns what was kept and what is missing.
| Argument | Type | Description |
|---|---|---|
| done | boolean | |
| task* | string | The task id geothought_next_document gave. |
| model | string | |
| token* | string | The lease token that came with it. |
| replies* | object |
{
"type": "object",
"required": [
"task",
"token",
"replies"
],
"properties": {
"done": {
"type": "boolean",
"default": false
},
"task": {
"type": "string",
"description": "The task id geothought_next_document gave."
},
"model": {
"type": "string"
},
"token": {
"type": "string",
"description": "The lease token that came with it."
},
"replies": {
"type": "object"
}
}
}Use to list GeoThought's ready-made schemas (refs t:<id>), which a project can read with as they are. Returns each one's ref, name, kind, description and question count.
No arguments.
{
"type": "object",
"properties": {}
}Use to change one of the user's projects: name, description, schemas (refs; replaces the list), corpora, period. Making it public is geothought_publish_project. Returns the project's new settings.
| Argument | Type | Description |
|---|---|---|
| name | string | |
| corpora | array | |
| project* | string | |
| schemas | array | |
| year_to | integer | null | |
| year_from | integer | null | |
| description | string |
{
"type": "object",
"required": [
"project"
],
"properties": {
"name": {
"type": "string"
},
"corpora": {
"type": "array",
"items": {
"type": "string"
}
},
"project": {
"type": "string"
},
"schemas": {
"type": "array",
"items": {
"type": "string"
}
},
"year_to": {
"type": [
"integer",
"null"
]
},
"year_from": {
"type": [
"integer",
"null"
]
},
"description": {
"type": "string"
}
}
}Use to list the user's own document files (every PDF, Word or PowerPoint file in their data library, however it was uploaded). Returns how far each has been read (state; steps: text, layout, OCR), a short reason when a step failed, the projects holding it, and how many of their files their projects hold against GeoThought's reading limit. New files are uploaded with upload_data.
No arguments.
{
"type": "object",
"properties": {}
}Use to read one data catalogue entry by its slug (from search_data_catalogue), e.g. before citing or downloading it. Returns its fields, coverage, period, citation and attribution, licence, the tables behind it, how to download it and an API example, and recent changes.
| Argument | Type | Description |
|---|---|---|
| slug* | string |
{
"type": "object",
"required": [
"slug"
],
"properties": {
"slug": {
"type": "string"
}
}
}Use to read one HyLogger-scanned hole by its key (from find_core_scans). Returns where and what it is, the scanned depth range, logged geology per metre (summarised to up to 15 rock types and their metres), the spectral datasets and which were kept, the data-quality issues found and their depths, the core-type model's summary, sample intervals, and links to its CoreFlagging page, line-scan image and sample sheets.
| Argument | Type | Description |
|---|---|---|
| key* | string |
{
"type": "object",
"required": [
"key"
],
"properties": {
"key": {
"type": "string"
}
}
}Use to read one feature of a published data-library layer (`layer_id` and `feature_id` from query_features). Returns all of its attributes and its GeoJSON geometry.
| Argument | Type | Description |
|---|---|---|
| layer_id* | string | |
| feature_id* | integer |
{
"type": "object",
"required": [
"layer_id",
"feature_id"
],
"properties": {
"layer_id": {
"type": "string"
},
"feature_id": {
"type": "integer"
}
}
}Use when the user asks to download a dataset's original uploaded file. Returns a short-lived download link. Available to the dataset's owner, and to anyone when the owner allowed original download of a public dataset.
| Argument | Type | Description |
|---|---|---|
| dataset_id* | string |
{
"type": "object",
"required": [
"dataset_id"
],
"properties": {
"dataset_id": {
"type": "string"
}
}
}Use when the user asks for the drillholes in an area with their downhole paths or gold grades. Give a bbox (west/south/east/north) or a point (lon/lat, with radius_km). Returns holes[], each with collar lon/lat, src (company or openfile) and intervals[{f, t, dip, az, au}] desurveyed from dip and azimuth (company holes carry per-interval Au grade; open-file holes have collar and total depth only), capped for size; plus a mapUrl and view3dUrl, a link to the platform's 3D view of the same area (grades by element, rock log, sections; opens in a browser, no sign-in).
| Argument | Type | Description |
|---|---|---|
| lat | number | Point mode: centre latitude (with lon). |
| lon | number | Point mode: centre longitude (with lat). |
| east | number | |
| west | number | |
| north | number | |
| south | number | |
| radius_km | number | Point mode: half-size of the square, km (0.5-8). |
{
"type": "object",
"required": [],
"properties": {
"lat": {
"type": "number",
"description": "Point mode: centre latitude (with lon)."
},
"lon": {
"type": "number",
"description": "Point mode: centre longitude (with lat)."
},
"east": {
"type": "number"
},
"west": {
"type": "number"
},
"north": {
"type": "number"
},
"south": {
"type": "number"
},
"radius_km": {
"type": "number",
"default": 3,
"description": "Point mode: half-size of the square, km (0.5-8)."
}
}
}Use when the user asks for the assay geochemistry (sample concentrations) around a point. Returns, per sample type (STREA stream sediment, ROCKC rock chip, DRILL, SHALL, SOIL) and per element, concentration statistics in ppm (n, min, p50, p90, max, mean), and the highest-value samples, with coordinates, for the first element in `elements`. Pass raw=true for the per-sample arrays when there are 500 samples or fewer.
| Argument | Type | Description |
|---|---|---|
| lat* | number | |
| lon* | number | |
| raw | boolean | Return the per-sample arrays if n <= 500. |
| elements | array | Subset of Cu, Au, Mo, Ag, Pb, Zn, Co, Bi, Ni, W, As, Sn, Cr, Sb (default Au, Cu, As). The highest-value samples are ranked by the first. |
| radius_km | number | 0.5-25 km |
| sample_types | array | Subset of STREA, ROCKC, DRILL, SHALL, SOIL (default all). |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"raw": {
"type": "boolean",
"description": "Return the per-sample arrays if n <= 500."
},
"elements": {
"type": "array",
"items": {
"type": "string"
},
"description": "Subset of Cu, Au, Mo, Ag, Pb, Zn, Co, Bi, Ni, W, As, Sn, Cr, Sb (default Au, Cu, As). The highest-value samples are ranked by the first."
},
"radius_km": {
"type": "number",
"default": 5,
"description": "0.5-25 km"
},
"sample_types": {
"type": "array",
"items": {
"type": "string"
},
"description": "Subset of STREA, ROCKC, DRILL, SHALL, SOIL (default all)."
}
}
}Use when the user asks what the geology is at a point in Western Australia. Returns the GSWA 1:500k unit and its place in its 1:250k sheet's stratigraphic column (map2loop: above, below, nearest contact and its type), the nearest fault (name, length, azimuth, the units either side), distance to geological features, petrophysics, and the litho2strat stratigraphy. Tenure, mines, holes and geochemistry around a point come from inspect_point.
| Argument | Type | Description |
|---|---|---|
| lat* | number | |
| lon* | number |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
}
}
}Use to read the details of an image dataset (`image_id` from list_datasets). Returns its size, format, colour mode and frame count; for the image's owner, also its camera metadata (EXIF) without the artist field.
| Argument | Type | Description |
|---|---|---|
| image_id* | string |
{
"type": "object",
"required": [
"image_id"
],
"properties": {
"image_id": {
"type": "string"
}
}
}Use to check on a job queued with submit_job. Give its taskId; returns its status (pending, running, done or failed) with the result once done, or a short error message if it failed. A job queued by another user reads as not found.
| Argument | Type | Description |
|---|---|---|
| taskId* | string |
{
"type": "object",
"required": [
"taskId"
],
"properties": {
"taskId": {
"type": "string"
}
}
}Use when the user asks for the geophysical or structural values the prospectivity model reads at specific points. For up to 25 points returns geophysics (magnetic and gravity intensity and derivatives) and structure (distances to faults and to magnetic and gravity worms, fault density); null means the point is off the layer or has no data.
| Argument | Type | Description |
|---|---|---|
| groups | array | Default both. |
| points* | array | 1-25 points |
{
"type": "object",
"required": [
"points"
],
"properties": {
"groups": {
"type": "array",
"items": {
"enum": [
"geophysics",
"structure"
],
"type": "string"
},
"description": "Default both."
},
"points": {
"type": "array",
"items": {
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
}
}
},
"description": "1-25 points"
}
}
}Use when the user asks for field photographs of outcrops or rock samples near a point in Western Australia (GSWA's WAROX field database). Returns, for the nearest WAROX sites and samples that have photographs, their description and each photo's address and caption, served by DMIRS (waroxphotos.demirs.wa.gov.au; credit "Photo: GSWA WAROX field database (DMIRS), CC BY 4.0").
| Argument | Type | Description |
|---|---|---|
| lat* | number | |
| lon* | number | |
| sites | integer | How many sites to look up. |
| radius_km | number | At most 20 km. |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"sites": {
"type": "integer",
"default": 5,
"maximum": 15,
"description": "How many sites to look up."
},
"radius_km": {
"type": "number",
"default": 2,
"description": "At most 20 km."
}
}
}Use when the user asks for an overview of a 1:250k map area ('areas'), a geological province ('provinces'), a commodity ('commodities') or a Minedex mine or deposit ('mines'), built from the platform's own data. Without slug: returns the summaries of that kind whose name matches q, with their report and Minedex counts. With slug: returns that summary's figures (reports by decade, commodities and operators by report count, Minedex sites, training sites...) and its sources.
| Argument | Type | Description |
|---|---|---|
| q | string | Part of a name or code (e.g. "kalgoorlie", "SH51-09", "gold"). |
| kind* | string | |
| slug | string | A page's slug from an earlier answer. |
| limit | integer |
{
"type": "object",
"required": [
"kind"
],
"properties": {
"q": {
"type": "string",
"description": "Part of a name or code (e.g. \"kalgoorlie\", \"SH51-09\", \"gold\")."
},
"kind": {
"enum": [
"areas",
"provinces",
"commodities",
"mines"
],
"type": "string"
},
"slug": {
"type": "string",
"description": "A page's slug from an earlier answer."
},
"limit": {
"type": "integer",
"default": 20,
"maximum": 100
}
}
}Use to see what one WA data library layer (a "lib:..." id from search_wa_data_library) holds around a point, e.g. the bedrock units, dykes, heritage places or WAROX observations around a prospect. Returns the features within the radius, sorted by distance, nearest first, each with its fields, label, distance and a point on it.
| Argument | Type | Description |
|---|---|---|
| lat* | number | |
| lon* | number | |
| layer* | string | |
| limit | integer | |
| radius_km | number | At most 50 km. |
{
"type": "object",
"required": [
"layer",
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"layer": {
"type": "string"
},
"limit": {
"type": "integer",
"default": 50,
"maximum": 200
},
"radius_km": {
"type": "number",
"default": 5,
"description": "At most 50 km."
}
}
}Use when the user asks what is known about the ground around a point in Western Australia: who holds it, what has been mined, found or drilled nearby, and what limits access. Returns a compact summary within `radius_km` of: tenure (whether the point is under a live tenement and its holder, from the public tenement register), nearby mines (MINEDEX) and known deposits, open-file drillholes, company drillhole and surface geochemistry (maximum Au and the highest-grade downhole intercept), GSWA regional geochemistry, WAMEX reports (each with an anumber, which wamex_report_files or document_files turn into files), the nearest mapped fault and magnetic and gravity worms, past exploration activity, host geology and regolith, land access (conservation reserves, native title, ILUAs, pastoral leases) and indicative Aboriginal heritage, plus a mapUrl. Pass `sections` to get those sections in full detail.
| Argument | Type | Description |
|---|---|---|
| lat* | number | Latitude (WA, ~-35 to -14) |
| lon* | number | Longitude (WA, ~113-129) |
| metal | string | Optional; filters known deposits to this commodity (Au, Cu, Ni, W, Sn, Co, Ta, Mn, REE) |
| sections | array | Sections to return in full detail; omit for the compact summary only. geochem = company drillhole/surface + GSWA assay anomalies; geology = host bedrock + regolith cover; landaccess = conservation reserves, native title, Indigenous Land Use Agreements (ILUAs) & pastoral leases; heritage = indicative Aboriginal heritage (not a clearance). |
| radius_km | number | Search radius, 1-50 km |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number",
"description": "Latitude (WA, ~-35 to -14)"
},
"lon": {
"type": "number",
"description": "Longitude (WA, ~113-129)"
},
"metal": {
"type": "string",
"description": "Optional; filters known deposits to this commodity (Au, Cu, Ni, W, Sn, Co, Ta, Mn, REE)"
},
"sections": {
"type": "array",
"items": {
"enum": [
"tenure",
"mines",
"deposits",
"drilling",
"geochem",
"reports",
"activity",
"structure",
"geology",
"landaccess",
"heritage"
],
"type": "string"
},
"description": "Sections to return in full detail; omit for the compact summary only. geochem = company drillhole/surface + GSWA assay anomalies; geology = host bedrock + regolith cover; landaccess = conservation reserves, native title, Indigenous Land Use Agreements (ILUAs) & pastoral leases; heritage = indicative Aboriginal heritage (not a clearance)."
},
"radius_km": {
"type": "number",
"default": 10,
"description": "Search radius, 1-50 km"
}
}
}Use when the user asks for an overview of what is inside a rectangular area (a bounding box). Returns counts for the area: tenements (live, pending, dead), mines (MINEDEX), known deposits, open-file drillholes, company drillhole and surface geochemistry and GSWA geochemistry (count and maximum Au), WAMEX reports, past exploration activity and land-access constraints (conservation reserves, native title, pastoral leases), plus a mapUrl. export_data with the same bbox returns the rows behind a count.
| Argument | Type | Description |
|---|---|---|
| east* | number | Max longitude |
| west* | number | Min longitude (WA ~113-129) |
| metal | string | Optional; filters known deposits to this commodity (Au, Cu, Ni, W, Sn, Co, Ta, Mn, REE) |
| north* | number | Max latitude |
| south* | number | Min latitude (WA ~-35 to -14) |
{
"type": "object",
"required": [
"west",
"south",
"east",
"north"
],
"properties": {
"east": {
"type": "number",
"description": "Max longitude"
},
"west": {
"type": "number",
"description": "Min longitude (WA ~113-129)"
},
"metal": {
"type": "string",
"description": "Optional; filters known deposits to this commodity (Au, Cu, Ni, W, Sn, Co, Ta, Mn, REE)"
},
"north": {
"type": "number",
"description": "Max latitude"
},
"south": {
"type": "number",
"description": "Min latitude (WA ~-35 to -14)"
}
}
}Use when the user asks which datasets they can read in the data library: their own uploads, datasets shared with them and datasets published publicly. Optional `query` filters by name or description. Returns each dataset's id, name, description, visibility, whether the user owns it and their access, its kind (spatial, table, samples, raster, document, text or image), the published layer (id, geometry type, feature count, fields), table (id, row count, columns), document (id, title, pages; text files are documents too), raster (id, size, bands) or image (id, size), its licence and projects, and the latest version's state.
| Argument | Type | Description |
|---|---|---|
| query | string | Case-insensitive substring of the name or description. |
| scope | string | all: everything you may read; mine: your own; shared: shared with you by others; public: published by anyone. |
| project | string | A project id (see `projects` in each entry): only that project's datasets. |
{
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "Case-insensitive substring of the name or description."
},
"scope": {
"enum": [
"all",
"mine",
"shared",
"public"
],
"type": "string",
"default": "all",
"description": "all: everything you may read; mine: your own; shared: shared with you by others; public: published by anyone."
},
"project": {
"type": "string",
"description": "A project id (see `projects` in each entry): only that project's datasets."
}
}
}Use when the user wants to know which data layers export_data can export. Returns each layer with its filters, row cap and whether a bbox is required. The layers are tenements, minedex, drillholes, wamex, activity (historical exploration), prospectivity (score grid), sites (training deposits), faults, worms_mag and worms_grav. The database holds many more tables (full text of the WAMEX reports, the raw MDHDB assay mirror; 5,000+ tables in all): sql_tables lists them and sql_query reads them.
No arguments.
{
"type": "object",
"properties": {}
}Use to find a public geophysics or terrain grid to read values from: GA's national gravity, magnetics (TMI, RTP, 1VD...), radiometrics, isostatic residuals, SGTool products and DEM, and GSWA's WA Moho and scanned geological map mosaics. Returns each grid's raster_id, family, name, title, bounds, colour stretch and unit. Pass the raster_id to sample_raster or describe_raster.
| Argument | Type | Description |
|---|---|---|
| q | string | Words in the title, name or family (e.g. "moho", "rtp"); empty lists all. |
{
"type": "object",
"properties": {
"q": {
"type": "string",
"description": "Words in the title, name or family (e.g. \"moho\", \"rtp\"); empty lists all."
}
}
}Use when the user asks for a list of known mineral sites, deposits or occurrences. Returns up to 200 sites, each with name, commodity, label, isPositive (whether the model uses it as a confirmed training example) and coordinates. Filter by metal and/or a bounding box; positive_only keeps confirmed occurrences only.
| Argument | Type | Description |
|---|---|---|
| bbox | array | [west, south, east, north] lon/lat filter (optional). |
| metal | string | Au, Cu, Ni, W, Sn, Co, Ta, Mn, REE |
| positive_only | boolean |
{
"type": "object",
"required": [],
"properties": {
"bbox": {
"type": "array",
"items": {
"type": "number"
},
"description": "[west, south, east, north] lon/lat filter (optional)."
},
"metal": {
"type": "string",
"description": "Au, Cu, Ni, W, Sn, Co, Ta, Mn, REE"
},
"positive_only": {
"type": "boolean"
}
}
}Use when the user asks which metals can be scored. Returns the metals the prospectivity model supports.
No arguments.
{
"type": "object",
"properties": {}
}Use when the user asks for a vertical cross-section through a drillhole. The section runs along the hole's own azimuth (east-west for a vertical hole or one whose direction was not recorded), as far as the drilling near the line reaches. Returns the line (ends, azimuth, bearing, which way it looks, length, the projected CRS), the 1:500k bedrock units it crosses with their distances along it, the faults it crosses with their apparent dip in the section, the local strike, and the holes within width_m of the line: their distance along it and off it, collar RL, depth and up to three intercepts for the element. Give the hole's collar (lon/lat, e.g. from find_drillholes) and, when several holes stand close together, its name.
| Argument | Type | Description |
|---|---|---|
| lat* | number | |
| lon* | number | |
| hole | string | The hole's name; else the hole nearest lon/lat. |
| element | string | |
| width_m | number | Holes this far either side of the line are on the section. |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
},
"hole": {
"type": "string",
"description": "The hole's name; else the hole nearest lon/lat."
},
"element": {
"enum": [
"Ag",
"As",
"Au",
"Ba",
"Be",
"Bi",
"Co",
"Cr",
"Cs",
"Cu",
"Fe",
"Li",
"Mn",
"Mo",
"Nb",
"Ni",
"P",
"Pb",
"Sb",
"Sn",
"Sr",
"Th",
"U",
"W",
"Y",
"Zn"
],
"type": "string",
"default": "Au"
},
"width_m": {
"enum": [
25,
50,
100,
250
],
"type": "number",
"default": 50,
"description": "Holes this far either side of the line are on the section."
}
}
}Use to give the user a link that opens the public data map (https://geo.nlp-tlp.org/map) on a view that shows an answer. The view can set layers to turn on (map layer ids such as drillholes, tenements, wamex, geology_au, geology, faults, offshore_titles / offshore_titles_pending (NOPTA), ilua (NNTT ILUAs), or library ids "lib:..."), a company or project's holdings drawn over the map, the drillhole filter, a place to centre on, and a pick (a point and radius, which opens the point's summary card; with view 3d, the 3D drillhole view, at most 10 km, optionally on one hole's card or its section). Returns the link; it builds the address only and fetches nothing. The map page itself can print the view or save it as PNG or PDF; no tool does that.
| Argument | Type | Description |
|---|---|---|
| bbox | array | [w, s, e, n]: picks the rectangle (its summary card, or 3D with view 3d). |
| hole | string | With view 3d: open this hole's card (its name, e.g. MFED024). |
| pick | array | [lon, lat] |
| view | string | |
| types | array | |
| center | array | [lon, lat] or [lon, lat, zoom] (zoom 4 = the State, 8 = a region, 12 = a mine). |
| layers | array | |
| company | string | |
| element | string | With view 3d: colour the holes by this element's grade (e.g. Au). |
| project | string | |
| section | boolean | With view 3d and hole: open the section through the hole. |
| radius_km | number | |
| min_depth_m | number |
{
"type": "object",
"properties": {
"bbox": {
"type": "array",
"items": {
"type": "number"
},
"maxItems": 4,
"minItems": 4,
"description": "[w, s, e, n]: picks the rectangle (its summary card, or 3D with view 3d)."
},
"hole": {
"type": "string",
"description": "With view 3d: open this hole's card (its name, e.g. MFED024)."
},
"pick": {
"type": "array",
"items": {
"type": "number"
},
"maxItems": 2,
"minItems": 2,
"description": "[lon, lat]"
},
"view": {
"enum": [
"map",
"3d"
],
"type": "string",
"default": "map"
},
"types": {
"type": "array",
"items": {
"enum": [
"dd",
"rc",
"ac",
"shallow",
"other"
],
"type": "string"
}
},
"center": {
"type": "array",
"items": {
"type": "number"
},
"maxItems": 3,
"minItems": 2,
"description": "[lon, lat] or [lon, lat, zoom] (zoom 4 = the State, 8 = a region, 12 = a mine)."
},
"layers": {
"type": "array",
"items": {
"type": "string"
}
},
"company": {
"type": "string"
},
"element": {
"enum": [
"Ag",
"As",
"Au",
"Ba",
"Be",
"Bi",
"Co",
"Cr",
"Cs",
"Cu",
"Fe",
"Li",
"Mn",
"Mo",
"Nb",
"Ni",
"P",
"Pb",
"Sb",
"Sn",
"Sr",
"Th",
"U",
"W",
"Y",
"Zn"
],
"type": "string",
"description": "With view 3d: colour the holes by this element's grade (e.g. Au)."
},
"project": {
"type": "string"
},
"section": {
"type": "boolean",
"description": "With view 3d and hole: open the section through the hole."
},
"radius_km": {
"type": "number"
},
"min_depth_m": {
"type": "number"
}
}
}Use when the user asks how long a line is or how large an area is. Give points as [[lon, lat], ...]; closed=true reads them as a polygon (the first point need not be repeated). Returns the geodesic (WGS84 ellipsoid) length of a line in km with each segment and its bearing, or the area (km² and ha) and perimeter of a polygon.
| Argument | Type | Description |
|---|---|---|
| closed | boolean | true: a polygon (area); false: a line. |
| points* | array |
{
"type": "object",
"required": [
"points"
],
"properties": {
"closed": {
"type": "boolean",
"default": false,
"description": "true: a polygon (area); false: a line."
},
"points": {
"type": "array",
"items": {
"type": "array",
"items": {
"type": "number"
},
"maxItems": 2,
"minItems": 2
},
"maxItems": 500,
"minItems": 2
}
}
}Use only after the user has agreed to the size that estimate_download reported, to package those files for the user to download. The package is built in the background as ZIP parts of about 10 GB, each listing its files (MANIFEST.csv) and any that could not be included (MISSING.csv). `include`: `all` (default) or `pdf` (report PDFs only, often much smaller). Only the signed-in user receives it: one email to their own address when it is ready, with the download links; it is kept for 14 days, then removed. Nothing else is changed. One package per user at a time; very large ones (over 200 GB) wait for the platform's approval. Returns the package's id and state; follow it with download_status.
| Argument | Type | Description |
|---|---|---|
| include | string | |
| estimate_id* | string |
{
"type": "object",
"required": [
"estimate_id"
],
"properties": {
"include": {
"enum": [
"all",
"pdf"
],
"type": "string",
"default": "all"
},
"estimate_id": {
"type": "string"
}
}
}Use to read the features or rows of a data-library layer (`layer_id` from describe_dataset), a WA data library layer (`layer_id` "lib:..." from search_wa_data_library, e.g. the geology polygons in a box or of one rock type) or a table (`table_id`). Optional `bbox` (minx,miny,maxx,maxy in WGS84 degrees), `filters` (list of {field, op, value}; op is eq/ne/lt/lte/gt/gte for numeric fields, eq/ne/contains for text), `fields` (attribute names to return), `limit` (at most 1000; 200 with geometry on a library layer) and `cursor` from a previous page. Set `geometry` true to include GeoJSON geometry. Returns the features with their attributes; a library layer also gives the count of all matches and its field names.
| Argument | Type | Description |
|---|---|---|
| bbox | string | |
| limit | integer | |
| cursor | string | |
| fields | array | |
| filters | array | |
| geometry | boolean | |
| layer_id | string | |
| table_id | string | For table datasets (CSV / XLSX), in place of layer_id. |
{
"type": "object",
"properties": {
"bbox": {
"type": "string"
},
"limit": {
"type": "integer",
"default": 100,
"maximum": 1000,
"minimum": 1
},
"cursor": {
"type": "string"
},
"fields": {
"type": "array",
"items": {
"type": "string"
}
},
"filters": {
"type": "array",
"items": {
"type": "object",
"required": [
"field",
"op"
],
"properties": {
"op": {
"type": "string"
},
"field": {
"type": "string"
},
"value": {}
}
}
},
"geometry": {
"type": "boolean",
"default": false
},
"layer_id": {
"type": "string"
},
"table_id": {
"type": "string",
"description": "For table datasets (CSV / XLSX), in place of layer_id."
}
}
}Use to read one page of a data-library document (`document_id` from list_datasets or search_dataset_documents; `page` is 1-based). Returns the page's text.
| Argument | Type | Description |
|---|---|---|
| page* | integer | |
| document_id* | string |
{
"type": "object",
"required": [
"document_id",
"page"
],
"properties": {
"page": {
"type": "integer",
"minimum": 1
},
"document_id": {
"type": "string"
}
}
}Use to read a raster's values at up to 50 WGS84 points ({lon, lat}). Returns the value of each band at each point; a point outside the raster has no value.
| Argument | Type | Description |
|---|---|---|
| points* | array | |
| raster_id* | string |
{
"type": "object",
"required": [
"raster_id",
"points"
],
"properties": {
"points": {
"type": "array",
"items": {
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
}
}
},
"maxItems": 50,
"minItems": 1
},
"raster_id": {
"type": "string"
}
}
}Use when the user asks how prospective a location in Western Australia is for a metal (Au, Cu, Ni, W, Sn, Co, Ta, Mn). Returns the model's prospectivity score (%), tier, data coverage and uncertainty, the nearest known mineral site, the geochemical and geophysical signals behind the score with each sub-model's score and weight, and a mapUrl that opens the point on the web map.
| Argument | Type | Description |
|---|---|---|
| lat* | number | Latitude (WA, ~-35 to -14) |
| lon* | number | Longitude (WA, ~113-129) |
| metal | string |
{
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number",
"description": "Latitude (WA, ~-35 to -14)"
},
"lon": {
"type": "number",
"description": "Longitude (WA, ~113-129)"
},
"metal": {
"type": "string",
"default": "Au"
}
}
}Use when the user asks how prospective an area drawn as a polygon is for a metal. Scores an 8x8 grid of points inside the polygon and returns the median score (%), tier, the lowest and highest grid score, uncertainty and a mapUrl. A large or data-dense polygon can take longer than a minute; submit_job runs it in the background.
| Argument | Type | Description |
|---|---|---|
| metal | string | |
| points* | array | Polygon ring |
{
"type": "object",
"required": [
"points"
],
"properties": {
"metal": {
"type": "string",
"default": "Au"
},
"points": {
"type": "array",
"items": {
"type": "object",
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
}
}
},
"description": "Polygon ring"
}
}
}Use when the user asks how prospective the ground is along a road, track or planned drive. Give the ordered vertices of the route (for example a Mapbox Directions geometry); the line is resampled every `spacing_km` and each sample is scored. Returns the route length, sample count, highest and median score, the distance along the route of the highest score, and the score peaks (hotspots), each with its distance along the route, score, tier, nearest known site and a mapUrl.
| Argument | Type | Description |
|---|---|---|
| metal | string | |
| points* | array | Ordered route vertices (>=2), start to end |
| spacing_km | number | Sample interval along the route, km (0.2-25) |
{
"type": "object",
"required": [
"points"
],
"properties": {
"metal": {
"type": "string",
"default": "Au"
},
"points": {
"type": "array",
"items": {
"type": "object",
"required": [
"lon",
"lat"
],
"properties": {
"lat": {
"type": "number"
},
"lon": {
"type": "number"
}
}
},
"description": "Ordered route vertices (>=2), start to end"
},
"spacing_km": {
"type": "number",
"default": 1,
"description": "Sample interval along the route, km (0.2-25)"
}
}
}Use when the user asks what data the platform holds on a subject, or where a dataset comes from and under which licence. Searches the platform's data catalogue (geo.nlp-tlp.org/data). Returns each dataset's title, summary, family, coverage, source, licence (and whether commercial use is allowed), row count, last update, its catalogue page, and the map layer ids it draws (make_map_link shows them). The ~600 DMIRS DASC packages of the WA data library are searched with search_wa_data_library.
| Argument | Type | Description |
|---|---|---|
| q | string | Words in a title or summary; empty lists everything. |
| limit | integer | |
| family | string | A family key from an earlier answer (e.g. geophysics, tenure). |
{
"type": "object",
"properties": {
"q": {
"type": "string",
"description": "Words in a title or summary; empty lists everything."
},
"limit": {
"type": "integer",
"default": 20,
"maximum": 100
},
"family": {
"type": "string",
"description": "A family key from an earlier answer (e.g. geophysics, tenure)."
}
}
}Use to find words in the PDF documents of the data library that the user can read. Accepts Postgres websearch syntax (quoted phrases, OR, -excluded). Returns each hit's document id, dataset, page and a snippet with <mark> around the matches. Optional `dataset_ids` limits the search to those datasets.
| Argument | Type | Description |
|---|---|---|
| limit | integer | |
| query* | string | |
| dataset_ids | array |
{
"type": "object",
"required": [
"query"
],
"properties": {
"limit": {
"type": "integer",
"default": 20,
"maximum": 100,
"minimum": 1
},
"query": {
"type": "string"
},
"dataset_ids": {
"type": "array",
"items": {
"type": "string"
}
}
}
}Use when the user asks what exploration reports, geological survey publications or research papers say about a topic, place, company or mineral. Read-only full-text search of public documents, one corpus at a time; it changes nothing. Returns ranked passages, each with the document's key, title and series, the page, the matching sentence, its position on the page and a layout label (text, table, figure, ...); pass a hit's corpus and doc_key to document_files for its files. Corpora: 'wamex', Western Australia's open-file company exploration reports (what a company did on a tenement); 'sarig', South Australia's; 'ebookshop', the Geological Survey of WA's publications (WA geology, stratigraphy, mineral systems; the default); 'hylogger', drill-core spectral reports for every State (what minerals scanned core contains); 'geolit', peer-reviewed geoscience papers worldwide (methods, mechanisms, analogues). Use plain words ('banded iron formation'), not operators. `labels` limits hits to tables, figures, captions, headings and so on in any corpus. A geolit title may have been read off the first page; its `series` (journal and year) is from the bibliography. `sampled` is true when a very common term ranks a sample rather than every hit.
| Argument | Type | Description |
|---|---|---|
| limit | integer | Hits to return (default 20, max 100). |
| query* | string | Plain words to look for. |
| corpus | string | Which corpus (default 'ebookshop'). |
| labels | array | Restrict to layout labels — 'table', 'title', 'plain text', 'figure', 'figure_caption'. Works on every corpus; canonical names cross all five. |
| min_ocr_conf | number | Drop low-confidence OCR text (0-1). On ebookshop this is a real threshold — 21% of OCR lines score under 0.5 and are mostly noise; sarig's OCR carries the same per-region confidence. wamex and geolit record no per-region confidence, so there it excludes OCR-derived regions instead of ranking them. Text-layer hits are never filtered. |
{
"type": "object",
"required": [
"query"
],
"properties": {
"limit": {
"type": "integer",
"description": "Hits to return (default 20, max 100)."
},
"query": {
"type": "string",
"description": "Plain words to look for."
},
"corpus": {
"enum": [
"wamex",
"ebookshop",
"geolit",
"hylogger",
"sarig"
],
"type": "string",
"description": "Which corpus (default 'ebookshop')."
},
"labels": {
"type": "array",
"items": {
"type": "string"
},
"description": "Restrict to layout labels — 'table', 'title', 'plain text', 'figure', 'figure_caption'. Works on every corpus; canonical names cross all five."
},
"min_ocr_conf": {
"type": "number",
"description": "Drop low-confidence OCR text (0-1). On ebookshop this is a real threshold — 21% of OCR lines score under 0.5 and are mostly noise; sarig's OCR carries the same per-region confidence. wamex and geolit record no per-region confidence, so there it excludes OCR-derived regions instead of ranking them. Text-layer hits are never filtered."
}
}
}Use when the user names a company or project and wants what it holds in Western Australia: its tenements, drillholes and WAMEX reports. A company groups its legal names ("Northern Star (KLV) Pty Ltd", "Northern Star Resources Ltd", ...) under one key, joint ventures counting for each party. Returns, closest name match first, each company's or project's key, name, counts (tenements and live ones, drillholes, reports), extent (bbox), the registered names it stands for, and a map link drawing its holdings. Pass the key to find_drillholes or make_map_link.
| Argument | Type | Description |
|---|---|---|
| q* | string | A company or project name, or part of one (3+ characters). |
| kind | string | Only companies or only projects. |
| limit | integer |
{
"type": "object",
"required": [
"q"
],
"properties": {
"q": {
"type": "string",
"description": "A company or project name, or part of one (3+ characters)."
},
"kind": {
"enum": [
"company",
"project"
],
"type": "string",
"description": "Only companies or only projects."
},
"limit": {
"type": "integer",
"default": 10,
"maximum": 30
}
}
}Use to find a Western Australian open-file dataset from the DMIRS Data and Software Centre (DASC) by its title: about 600 datasets (geology, geophysics compilations, mineralisation, WAROX field data, heritage, land, infrastructure, ...) and the 1:250 000 / 1:100 000 geological map sheets. Every word of q must appear in the title or category. Returns each dataset's title, group, category, last update, feature count and layer ids ("lib:...") for get_wa_library_features, query_features and make_map_link.
| Argument | Type | Description |
|---|---|---|
| q* | string | Words of the title, e.g. "dyke", "heritage", "KALGOORLIE". |
| limit | integer |
{
"type": "object",
"required": [
"q"
],
"properties": {
"q": {
"type": "string",
"description": "Words of the title, e.g. \"dyke\", \"heritage\", \"KALGOORLIE\"."
},
"limit": {
"type": "integer",
"default": 20,
"maximum": 50
}
}
}Use before writing a query against a table, to get its column names and types. Geometry columns are EPSG:4326 (lon/lat). `schema` is optional: without it the table is looked up by name (or pass "source.drillholes").
| Argument | Type | Description |
|---|---|---|
| table | string | The table's name, e.g. "drillholes" or "source.drillholes". |
| schema | string | Optional: public, source, derived or mdhdb_raw. |
{
"type": "object",
"properties": {
"table": {
"type": "string",
"description": "The table's name, e.g. \"drillholes\" or \"source.drillholes\"."
},
"schema": {
"type": "string",
"description": "Optional: public, source, derived or mdhdb_raw."
}
}
}Use when a question needs a custom query over the AgenticGeoMate geoscience database. Runs one read-only SELECT (or WITH/EXPLAIN) and returns the rows. Read-only is enforced by the database connection, not only checked: INSERT/UPDATE/DELETE/DDL cannot run. Tables can be joined across schemas (anumber links most tables to a WAMEX report; geom supports spatial joins via ST_DWithin). Rows are capped (default 1000, max 5000) and the query times out at 30s -- aggregate or add LIMIT for big tables. Results are paged: when page.next is set there are more rows -- call again with the same sql and page.next's limit and offset. A page is also cut to ~800k characters (wide rows): then truncated is true and page.next continues from the first row not returned. Account, token and chat tables are not readable. Read-only SQL against AgenticGeoMate's own PostgreSQL/PostGIS database (30-second timeout, up to 5,000 rows a page), documented at https://geo.nlp-tlp.org/docs#sql; sql_tables lists the tables and sql_describe a table's columns.
| Argument | Type | Description |
|---|---|---|
| sql* | string | A single SELECT statement, no trailing ';'. |
| limit | integer | Rows per page, max 5000 (default 1000). |
| offset | integer | Rows to skip: page.next.offset of the previous call (default 0). |
{
"type": "object",
"required": [
"sql"
],
"properties": {
"sql": {
"type": "string",
"description": "A single SELECT statement, no trailing ';'."
},
"limit": {
"type": "integer",
"description": "Rows per page, max 5000 (default 1000)."
},
"offset": {
"type": "integer",
"description": "Rows to skip: page.next.offset of the previous call (default 0)."
}
}
}Use before writing SQL, to see which tables exist when a question needs data the other tools do not cover. Returns every queryable table, grouped by schema, with estimated row counts. Schemas: public (GSWA assays, training sites, geophysics vectors), source (DMIRS distilled -- drillholes, tenements, geology, company_dh_intervals, plus 4800+ dasc_* mirror layers), derived (prospectivity grid), mdhdb_raw (the verbatim GSWA MDHDB, 60 gswa_* tables of raw per-element assays with lab methods).
No arguments.
{
"type": "object",
"properties": {}
}Use for calls that may take longer than a minute, such as score_region over a large or data-dense area. Queues the named tool with its arguments and returns a taskId at once; get_job returns the result when it is ready. Only the user who queued a job can read it.
| Argument | Type | Description |
|---|---|---|
| tool* | string | |
| arguments* | object |
{
"type": "object",
"required": [
"tool",
"arguments"
],
"properties": {
"tool": {
"enum": [
"score_point",
"score_region",
"score_route",
"top_prospects",
"inspect_point",
"compare_points",
"get_geochem",
"get_model_inputs_at_points"
],
"type": "string"
},
"arguments": {
"type": "object"
}
}
}Use when the user asks where the highest-scoring ground for a metal is across Western Australia. Returns up to `n` spaced-apart cells of the precomputed prospectivity grid, sorted by score, highest first, each with its coordinates and a mapUrl. Set exclude_nogo to leave out cells inside DBCA conservation reserves.
| Argument | Type | Description |
|---|---|---|
| n | integer | |
| metal | string | |
| exclude_nogo | boolean | Leave out cells inside DBCA conservation reserves. |
{
"type": "object",
"required": [],
"properties": {
"n": {
"type": "integer",
"default": 6
},
"metal": {
"type": "string",
"default": "Au"
},
"exclude_nogo": {
"type": "boolean",
"default": false,
"description": "Leave out cells inside DBCA conservation reserves."
}
}
}Use when the user asks to add a file from a public https link to their own data library (the platform downloads it; links to private or internal addresses are refused). Accepts any file the library takes: a PDF, Word or PowerPoint document, a spreadsheet or CSV, a GeoJSON, shapefile zip, GeoPackage, a GeoTIFF, an image. The dataset is private to the user and counts against their library's quota. Returns its dataset_id; call describe_dataset with it until its state is ready. A document gets its text read, then its layout (and OCR for scanned pages), and can be added to a GeoThought project with geothought_add_documents (uploads: [dataset_id]); geothought_uploads shows those steps.
| Argument | Type | Description |
|---|---|---|
| url* | string | A public https link to the file. |
| name | string | The dataset's name (default: the file's name in the link). |
| description | string |
{
"type": "object",
"required": [
"url"
],
"properties": {
"url": {
"type": "string",
"description": "A public https link to the file."
},
"name": {
"type": "string",
"description": "The dataset's name (default: the file's name in the link)."
},
"description": {
"type": "string"
}
}
}Use when the user wants the files of a WAMEX exploration report. Give an anumber (from inspect_point's reports, the wamex tables, or wamex_meta_* joins); returns mirror_files — the report's original PDF, appendix zips and an OCR'd copy, each with a presigned download_url (valid 7 days) — and the report abstract. A PDF that DMIRS shipped only inside one of the report's ZIPs is listed too, with extracted_from naming the ZIP and member. complete says whether the mirror holds every file DMIRS lists for the report; missing names each one it does not ({name, bytes, reason, detail}; reason unavailable_at_source when the department itself will not serve it).
| Argument | Type | Description |
|---|---|---|
| anumber* | string | Report number, e.g. 'A12345' or '12345'. |
{
"type": "object",
"required": [
"anumber"
],
"properties": {
"anumber": {
"type": "string",
"description": "Report number, e.g. 'A12345' or '12345'."
}
}
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 21, 2026.
Takes a password, API key or token as an argument, so the secret passes through the model's context and chat logs.
Fix: Read credentials from the server's own configuration (environment, OAuth) instead of a tool argument.