https://astreus.ai/api/v1; every
endpoint is documented in full in the
API Reference tab. {id} is a ticker,
SEC CIK or platform key (see Coverage).
The generated tool reference includes every operation and current
MCP mapping. CLI 0.6.0 can call every data operation through astreus api call <operation-id> even when no dedicated formatted command is listed below.
The JSON catalog carries exact schemas.
MCP tools may bound, filter or reshape REST data for licensing and context size;
a route’s existence does not imply unrestricted MCP redistribution.
Companies
Financial statements — page
Segments & KPIs — page
Filings & notes — page
Investor presentations — page
Transcripts & guidance — page
Events & news — page
Ownership & people — page
Macro & markets — page
Intelligence — page
Account
Transcript research filters
The Transcript Search workspace offers passage, company × topic, analyst-question, topic evolution, period-change, chronological timeline, company heatmap, side-by-side commentary, and regional map views. Heatmap cells count distinct matching calls by calendar quarter and open the underlying passages. The regional map groups matching calls or companies by issuer country recorded on each transcript, with region selection and a country list; it does not identify physical call venues or countries discussed. Missing locations remain visible as unknown. Click a country to filter its source evidence. Timeline and comparison counts describe loaded excerpts, with more evidence available through pagination. Empty heatmap/map cells do not establish transcript coverage or topic disappearance; broad sampled summaries hide these distributions. Transcript search accepts keywords (q), a company being discussed (mentioned), or an explicit lexical theme. Combine them to require a match in the same speaker passage. Theme matching expands to a curated set of phrases; it does not infer meaning from arbitrary natural-language questions.
Filter calls by company, a comma-separated companies basket of at most 20 identifiers, speaker name, role, section, or from/to dates. A basket and a single-company filter intersect. Pass includeContext=1 to return surrounding turns or the indexed Q&A exchange; context can include speakers outside the result filters and is flagged when shortened.
Combine two websearch expressions with q and q2; termMatch=all (default) requires both within the same speaker turn, while termMatch=any matches either. For example, q="Abu Dhabi"&q2=AI finds passages containing the phrase and AI. Region, subregion, and country refer to the issuer recorded on the transcript; city uses the company’s currently stored business-address city and has incomplete coverage. locationMode=outside excludes the intersection of selected geographic criteria. Unknown locations are excluded by default; includeUnknownLocation=1 explicitly includes them. The ENBD preset uses its platform identifier, XDFM000003, and country=US to search US issuers’ calls naming the bank. No matching evidence is a valid empty result.
Topic columns are editable in the matrix: add or rename topics, choose any/all matching phrases, add exclusions, reorder columns, or remove/re-add presets. The same definitions power company/topic, analyst Q&A, evolution, and period-change views. “Preview quote” and evidence drilldowns use the selected column’s matching rule. Named topic sets are saved in this browser for the signed-in account; “Save search” stores the columns together with the research scope. Column definitions also travel in the shared URL and can be exported as JSON.
The REST endpoints accept topicColumns, a JSON array of 1–12 definitions (id, label, terms, exclude, match), and column to apply one definition to a source-passage search. match=any combines matching phrases with OR, while match=all requires every phrase. Exclusions apply to the entire match, within the same speaker turn. Each phrase uses the existing English full-text index; phrases may undergo its normal stemming. Definitions are bounded for shareable URLs and validated by the API. Named sets do not currently sync across devices.
GET /transcripts/search/topics returns the company/topic matrix, calendar-quarter evolution, and counts for the current and immediately preceding equal-day comparison windows. Omitted from defaults to twelve months before to (today when omitted). The response includes scoped call/company coverage for both periods. Topics use inspectable lexical vocabularies and can overlap. Analyst-question views measure Q&A speaker turns and indexed exchanges, rather than assuming one turn equals one question. A cap of 5,000 passages per topic/period is explicit; the UI hides trend comparisons when capped. Cell ratios describe matching calls divided by calls in the filtered research scope, rather than global prevalence or a fixed comparison cohort. GET /transcripts/search/facets supplies region/subregion/country options; these options do not establish transcript coverage.
country: an uppercase ISO alpha-2 issuer country code, or unknown for missing/blank location metadata. Complete summaries expose countries with country, region, subregion, matching passage/call counts, and distinct company counts. This distribution includes all matching companies, rather than only the leading company chips.
sampled; narrow the search before drawing historical trend conclusions.
Additional research domains
- Calendars: earnings, live calls, economic, corporate and congressional events.
- Debt and supply chains: debt instruments, counterparties, ownership histories and management.
- Assets, FX and world data: asset attributes and histories, FX, regions, government entities, conflicts and organisations.
- Agent mode: the app-wide structured representation of existing pages, saved filters and account news monitors.