Octus

Search Octus Court Filings (tracked bankruptcy/litigation dockets) and return the most relevant excerpts. Scope either to one `company_id` (from find_company) or to a classification bucket (`industry_group` / `industry` / `region` numeric codes — see the `list_companies` tool for their legends) — never both. Narrow by a unix-second `start_date`/`end_date` window. Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_court_filings

Search Octus's private covenant corpus — entitlement-scoped deal documents (Covenant Analysis, Credit Agreements, OM/Indentures) for private transactions your account can access via Deal Term Analytics and documents in your FinDox data-room holdings. Access is resolved server-side from your entitlements — there is no access parameter, and an account with no matching entitlements gets an empty result with a note, never an error. A pure retriever: returns chunks and interpretation metadata, never a synthesized answer. Coverage starts 2024-01-01. Returns private-scoped documents only — no public documents, no overlap with search_covenants_public; pair the two tools for the full picture on a credit. Loan questions search credit agreements; bond questions search OM/indentures; both/unclear searches all three corpora — override with asset_type, or restrict to exact document types with `document_types`. Scope either to `company_ids` (from find_company) or to a classification bucket (`industry_group`/`industry`/`region` numeric codes — see the `list_companies` tool for their legends; resolved server-side to companies with covenant coverage, capped at 100) — never both. Leave both unset to search across all your entitled private documents — use search_mode=exhaustive to see which of your entitled credits match, then drill in with per-company direct calls. Returns `{filters, results, query_interpretation, analysis_hints, discovered_companies?, note?}` — `filters` echoes the EFFECTIVE scope searched (the `company_ids` you sent, OR the resolved industry_group/industry/region as `{code, name}` when scoped by classification instead) plus the date window; `results` is a FLAT list, one row per matching document (company_id, company_name, title, doc_type, execution_date, asset_type) with its excerpt chunks; `query_interpretation` reports how the query was read (mode, asset_type, reframed_query) so you can re-call with search_mode/asset_type overrides if it misread the question; `discovered_companies` (exhaustive mode) ranks companies by match count for follow-up per-company drill-down calls. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url`; never construct URLs.

search_covenants_private

Search Octus's public covenant corpus — analyst Covenant Analysis plus primary-source Credit Agreements (loans) and Offering Memorandum / Indenture documents (bonds) — and get back the most relevant excerpts with the metadata needed for covenant reasoning. A pure retriever: returns chunks and interpretation metadata, never a synthesized answer. Coverage starts 2024-01-01. Loan questions search credit agreements; bond questions search OM/indentures; both/unclear searches all three corpora — override with asset_type, or restrict to exact document types with `document_types`. Scope either to `company_ids` (from find_company) or to a classification bucket (`industry_group`/`industry`/`region` numeric codes — see the `list_companies` tool for their legends; resolved server-side to companies with covenant coverage, capped at 100) — never both. Leave both unset for corpus-wide discovery/screening — use search_mode=exhaustive to see which companies' documents match, then drill in with per-company direct calls. Not entitlement-scoped private deal documents — those are search_covenants_private (separate tool). Returns `{filters, results, query_interpretation, analysis_hints, discovered_companies?, note?}` — `filters` echoes the EFFECTIVE scope searched (the `company_ids` you sent, OR the resolved industry_group/industry/region as `{code, name}` when scoped by classification instead) plus the date window; `results` is a FLAT list, one row per matching document (company_id, company_name, title, doc_type, execution_date, asset_type) with its excerpt chunks; `query_interpretation` reports how the query was read (mode, asset_type, reframed_query) so you can re-call with search_mode/asset_type overrides if it misread the question; `discovered_companies` (exhaustive mode) ranks companies by match count for follow-up per-company drill-down calls. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url`; never construct URLs.

search_covenants_public

Search your FinDox deal-room documents (private deal docs: credit agreements, lender presentations, financials) and return the most relevant excerpts. Results always come from the deal-room holdings your account is entitled to; omit the scope filters to search all of them. Scope either to one `company_id` (from find_company) or to a classification bucket (`industry_group` / `industry` / `region` numeric codes — see the `list_companies` tool for their legends) — never both. Narrow by a unix-second `start_date`/`end_date` window. Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_data_rooms

Search Octus Fundamentals documents (earnings reports, presentations, memoranda) and return the most relevant excerpts. Scope either to one `company_id` (from find_company) or to a classification bucket (`industry_group` / `industry` / `region` numeric codes — see the `list_companies` tool for their legends) — never both. Time window is `num_periods` (fiscal-period lookback). Optionally restrict by `document_types` (1=Earnings Report, 2=Earnings Presentations, 3=Other Presentations and Reports, 4=Offerings, 5=Capital Structure, 6=Press Release, 7=Sustainability Reports, 8=Transcripts, 9=Compliance Certificates, 10=Transcript Summaries, 11=Private Company Analysis, 12=Earnings Flashes). Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_fundamentals

Search Octus Intel (editorial credit research/news) and return the most relevant excerpts. Scope either to one `company_id` (from find_company) or to a classification bucket (`industry_group` / `industry` / `region` numeric codes — see the `list_companies` tool for their legends) — never both. Narrow by a unix-second `start_date`/`end_date` window. Coverage starts 2013-01-01; a search that omits `start_date` defaults to roughly the last 5 years. Do not settle for the default when the question implies a window — set the dates from the question and from the company's `latest_intel_date` (get_company_details): a most-recent-update ask is served by a window anchored just before that date, a period-specific ask (a named year, quarter, or event era) by that period's own bounds, and anything older than the default window by an explicit `start_date` reaching back to 2013. For sector/market trend questions (not tied to one company), pass `intel_themes` instead — numeric theme codes, each selecting one Octus editorial theme feed (the 27 feeds are listed in the `intel_themes` parameter legend); `intel_themes` is a standalone scope (never combined with company_id/classification — use two separate calls for company + sector color). For 'what's trending / most popular intel' questions, set `sort_by=2` (popularity) to rank by reader engagement (no `question` needed) and choose a `timeframe` (1 = past 4 hours, 2 = past 24 hours (1 day), 3 = past 72 hours (3 days; the default), 4 = past 168 hours (1 week)); popularity combines with the same company/classification (or theme) scope, but not dates (use `timeframe`, not `start_date`/`end_date`). Theme feeds rarely rank among most-viewed articles, so prefer relevance mode for `intel_themes`. Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_intel

Search the documents you have uploaded to CreditAI and return the most relevant excerpts. Scope is automatic — always your own uploads; only files uploaded through the CreditAI web app are included (uploads elsewhere or via other Octus products are not). Optionally narrow within your uploads to documents naming one `company_id` (from find_company), and by a unix-second `start_date`/`end_date` window. Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_my_uploads

Search Octus Private Company Analysis reports and return the most relevant excerpts. Results always come from the documents your account is entitled to; omit the scope filters to search all of them. Scope either to one `company_id` (from find_company) or to a classification bucket (`industry_group` / `industry` / `region` numeric codes — see the `list_companies` tool for their legends) — never both. Narrow by a unix-second `start_date`/`end_date` window. Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_private_company_analysis

Search Octus SEC Filings and return the most relevant excerpts. Scope either to one `company_id` (from find_company) or to a classification bucket (`industry_group` / `industry` / `region` numeric codes — see the `list_companies` tool for their legends) — never both. Narrow by a unix-second `start_date`/`end_date` window, and optionally by `document_types` (1=8-K, 2=10-K, 3=10-Q, 4=proxy/14A). Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_sec_filings

Answer a natural-language `question` against a structured `datasource` over Octus data. Pick the `datasource` code and ask in plain English; the engine works out the data model, runs the query, and returns structured datasets (columns + rows) with the applied period/filters. Read-only; no company_id needed. `datasource` codes: • 1 (advisors) — advisor engagements over Octus Screener data: who advised which bankrupt / distressed companies (law firms, financial advisors, claims & noticing agents), appointment side (debtor vs. creditor / committee), roles, engagement fees, hourly-rate bands, and league tables by sector / jurisdiction / year, with engine-side grouping, binding-gated aggregation, and negative-existence screens. E.g. 'top financial advisors to debtors in 2023 retail bankruptcies by number of engagements', 'average partner hourly rate for lead debtor counsel'. • 2 (primary_tracker) — LevFin new-issue primary market: leveraged-loan and high-yield-bond tranches with launch / pricing dates, initial and final sizes, coupon / margin / yield / OID economics, price talk, ratings, bookrunners, sponsors, and use of proceeds, with engine-side counts, league tables, grouped averages / sums, and per-year / quarter time series. E.g. 'how many high-yield bonds priced in 2025', 'top bookrunners for 2025 leveraged loans', 'average TLB margin by market of risk'. (More datasource codes will be added as they productize.) Returns `datasets` — one entry per query the engine ran, each a `{label, columns, rows, ...}` table (rows are {column: value}) — plus an optional one-line `summary` and an optional `note`. When the question can't be answered (off-topic, a metric not in the model, or an aggregation the data can't support), the tool does not error — it returns an empty `datasets` list with a `note` explaining why in business terms, usually with a workable rephrasing; a partial answer likewise comes back with data plus a `note`. Always read the `note` and, when present, re-ask using its guidance. Rows are read-only structured values, not document excerpts — present the tables and any rankings directly. The response may include a one-row `results` index carrying the `url` of the Octus page this datasource's data lives on (a screen is a query, not a document, so individual rows have no links of their own) — when your answer uses this data, cite inline as a markdown link titled with that row's `title`, pointing at its `url`; never construct URLs.

search_structured_datasource

Search Octus earnings-call Transcripts and return the most relevant excerpts. Scope either to one `company_id` (from find_company) or to a classification bucket (`industry_group` / `industry` / `region` numeric codes — see the `list_companies` tool for their legends) — never both. Narrow by a unix-second `start_date`/`end_date` window. Returns `{product, filters, results}` — `filters` echoes the effective scope/window and `results` is a FLAT list, ONE ROW PER DOCUMENT (company_id, company_name, title, doc_type, period/date) with that document's excerpt `chunks`; rows are relevance-ordered and repeat their company, so several rows can share one. A row may carry a ready absolute `url` — when your answer uses its content, cite inline as a markdown link titled with the row's `title`, pointing at that `url` copied character-for-character; never construct URLs, and never shorten or truncate them (they are long by design).

search_transcripts