docket_query
ChatGPTPRIMARY tool — use this FIRST for any question about what NSEB filings say: facts, figures, evidence, positions, arguments, testimony, or where parties disagreed. Chunk-level hybrid (keyword + semantic) search over the full text of filings; returns passages with snippets, matter/exhibit/page metadata, and a url per result. Optionally scope by matter, topic, entity, or a single document (exhibit); with no scope filter set it searches across every proceeding. Call it before answering a factual question, and cite the matter/exhibit/page and url for each passage you use. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug/matter_number from the candidates, or omit the filter.
fetch
ChatGPTRead the full text of ONE NSEB passage by id. Use AFTER search returns a passage id. Returns {id, title, text, url, metadata} where metadata carries the document/matter/heading/page locators and url links to the source. When you quote it, cite the matter/exhibit/page and include the url.
fetch_window
ChatGPTRead a passage PLUS its neighbors in document order — use when a passage is cut off mid-answer (e.g. an IR response continuing past the passage fetch returned). Returns up to before preceding and after following passages from the SAME document, in order, each with its text; the response carries the document url for citation. Adjacent passages may repeat a few lines — dedup rather than double-quote.
get_document
ChatGPTGet metadata for ONE filing document by its doc_id — title, matter, tab, exhibit number, filing date, and a url to the document. Use after search or docket_query surface a doc_id. Every result carries a url to its DocketWatch source page — include that link when you cite it.
get_entity
ChatGPTGet ONE entity's detail — name, type, mention/matter counts, a url, and the matters it appears in (each with its own url), NEWEST FIRST. Use AFTER resolving the entity (via list_entities or docket_query) — e.g. for a party's track record across proceedings. Accepts a slug or a fuzzy display name. PAGED: returns up to limit matters (default 25, max 100); matter_count is the full total. When next_offset is present the list is partial — call again with that value as offset for the next page. NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug from the candidates, or omit the filter.
get_matter
ChatGPTGet ONE matter's full detail — metadata, tab counts, a url, and its document inventory (each document with its own url), NEWEST FILINGS FIRST. Use AFTER you've identified a matter (from docket_query or list_matters) and need its document list. Accepts a matter number (e.g. M12780) or a fuzzy title. PAGED: returns up to limit documents (default 25, max 100); document_count is the full total. When next_offset is present the inventory is partial — call again with that value as offset for the next page. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug/matter_number from the candidates, or omit the filter.
get_topic
ChatGPTGet ONE topic's detail — name, mention/matter counts, and a url. Use AFTER resolving the topic (via list_topics or docket_query). Accepts a slug or a fuzzy display name. NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug from the candidates, or omit the filter.
list_entities
ChatGPTBrowse or resolve DocketWatch entities (organizations and people in the filings) — name, matter count, and a url each. Use to FIND an entity slug (then scope docket_query by it, or call get_entity), not for content questions. Optional matter instead answers 'who is involved in this proceeding?' — returns that one matter's participants ranked by mention count (accepts a matter number or fuzzy title; unresolved values return candidates). NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Every result carries a url to its DocketWatch source page — include that link when you cite it.
list_matters
ChatGPTBrowse or find NSEB matters (proceedings) — returns matter number, title, status, document counts, and a url each, NEWEST FIRST. Use to DISCOVER or enumerate matters (e.g. 'how many matters', 'list open proceedings'), NOT to answer what a filing says — use docket_query for content. Optional q filters by title/number; status is open, closed, or all. PAGED: returns up to limit matters (default 25, max 100); matter_count is the full total. When next_offset is present the list is partial — call again with that value as offset to get the next page. Every result carries a url to its DocketWatch source page — include that link when you cite it.
list_topics
ChatGPTBrowse or resolve DocketWatch topics (regulatory subjects) — name, matter count, and a url each. Use to FIND a topic slug (then scope docket_query by it, or call get_topic), not to answer content questions. Optional matter instead answers 'what is this matter about?' — returns per-matter chunk/document counts for that one matter (accepts a matter number or fuzzy title; unresolved values return candidates). NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Every result carries a url to its DocketWatch source page — include that link when you cite it.
search
ChatGPTSearch Nova Scotia Energy Board (NSEB) regulatory filings for passages matching a query. Returns {id, title, url, snippet}: id is a passage id, title locates it ([matter] document — heading), url links to the source on DocketWatch, snippet is a readable excerpt — call fetch(id) for the full text. (For scoped or in-depth retrieval, docket_query is richer.) Prefer regulatory terminology; when you use a passage, cite its matter/exhibit/page and include the url.
docket_query
ClaudePRIMARY tool — use this FIRST for any question about what NSEB filings say: facts, figures, evidence, positions, arguments, testimony, or where parties disagreed. Chunk-level hybrid (keyword + semantic) search over the full text of filings; returns passages with snippets, matter/exhibit/page metadata, and a url per result. Optionally scope by matter, topic, entity, or a single document (exhibit); with no scope filter set it searches across every proceeding. Call it before answering a factual question, and cite the matter/exhibit/page and url for each passage you use. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug/matter_number from the candidates, or omit the filter.
fetch
ClaudeRead the full text of ONE NSEB passage by id. Use AFTER search returns a passage id. Returns {id, title, text, url, metadata} where metadata carries the document/matter/heading/page locators and url links to the source. When you quote it, cite the matter/exhibit/page and include the url.
fetch_window
ClaudeRead a passage PLUS its neighbors in document order — use when a passage is cut off mid-answer (e.g. an IR response continuing past the passage fetch returned). Returns up to before preceding and after following passages from the SAME document, in order, each with its text; the response carries the document url for citation. Adjacent passages may repeat a few lines — dedup rather than double-quote.
get_document
ClaudeGet metadata for ONE filing document by its doc_id — title, matter, tab, exhibit number, filing date, and a url to the document. Use after search or docket_query surface a doc_id. Every result carries a url to its DocketWatch source page — include that link when you cite it.
get_entity
ClaudeGet ONE entity's detail — name, type, mention/matter counts, a url, and the matters it appears in (each with its own url), NEWEST FIRST. Use AFTER resolving the entity (via list_entities or docket_query) — e.g. for a party's track record across proceedings. Accepts a slug or a fuzzy display name. PAGED: returns up to limit matters (default 25, max 100); matter_count is the full total. When next_offset is present the list is partial — call again with that value as offset for the next page. NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug from the candidates, or omit the filter.
get_matter
ClaudeGet ONE matter's full detail — metadata, tab counts, a url, and its document inventory (each document with its own url), NEWEST FILINGS FIRST. Use AFTER you've identified a matter (from docket_query or list_matters) and need its document list. Accepts a matter number (e.g. M12780) or a fuzzy title. PAGED: returns up to limit documents (default 25, max 100); document_count is the full total. When next_offset is present the inventory is partial — call again with that value as offset for the next page. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug/matter_number from the candidates, or omit the filter.
get_topic
ClaudeGet ONE topic's detail — name, mention/matter counts, and a url. Use AFTER resolving the topic (via list_topics or docket_query). Accepts a slug or a fuzzy display name. NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Filter values accept fuzzy names: if a value doesn't resolve you get ranked candidates instead of results — re-call with an exact slug from the candidates, or omit the filter.
list_entities
ClaudeBrowse or resolve DocketWatch entities (organizations and people in the filings) — name, matter count, and a url each. Use to FIND an entity slug (then scope docket_query by it, or call get_entity), not for content questions. Optional matter instead answers 'who is involved in this proceeding?' — returns that one matter's participants ranked by mention count (accepts a matter number or fuzzy title; unresolved values return candidates). NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Every result carries a url to its DocketWatch source page — include that link when you cite it.
list_matters
ClaudeBrowse or find NSEB matters (proceedings) — returns matter number, title, status, document counts, and a url each, NEWEST FIRST. Use to DISCOVER or enumerate matters (e.g. 'how many matters', 'list open proceedings'), NOT to answer what a filing says — use docket_query for content. Optional q filters by title/number; status is open, closed, or all. PAGED: returns up to limit matters (default 25, max 100); matter_count is the full total. When next_offset is present the list is partial — call again with that value as offset to get the next page. Every result carries a url to its DocketWatch source page — include that link when you cite it.
list_topics
ClaudeBrowse or resolve DocketWatch topics (regulatory subjects) — name, matter count, and a url each. Use to FIND a topic slug (then scope docket_query by it, or call get_topic), not to answer content questions. Optional matter instead answers 'what is this matter about?' — returns per-matter chunk/document counts for that one matter (accepts a matter number or fuzzy title; unresolved values return candidates). NOTE: links between topics/entities and documents are approximate and NOT comprehensive — use this tool to orient, never as proof that nothing else exists. For thorough retrieval use docket_query or search. Every result carries a url to its DocketWatch source page — include that link when you cite it.
search
ClaudeSearch Nova Scotia Energy Board (NSEB) regulatory filings for passages matching a query. Returns {id, title, url, snippet}: id is a passage id, title locates it ([matter] document — heading), url links to the source on DocketWatch, snippet is a readable excerpt — call fetch(id) for the full text. (For scoped or in-depth retrieval, docket_query is richer.) Prefer regulatory terminology; when you use a passage, cite its matter/exhibit/page and include the url.