Skip to main content
The News Feed endpoint aggregates articles from dozens of RSS sources across crypto, traditional finance, macro economics, geopolitics, research, and AI. Articles are scored by source tier (1-3) and enriched with coin tags for filtering.

Get news articles

GET /v1/news/feed Returns a paginated, story-deduplicated list of news articles ordered by publish date descending. Supports full-corpus title search plus filtering by category, coin tag, and date range. Each article gives the headline, source, publish time and a link to the original story. summary is always null: Sharpe does not store or serve publishers’ summaries, so open link to read the article. The field stays so existing clients keep working. coin_tags holds canonical asset ids (bitcoin); coin_symbols gives each one’s ticker in the same order (BTC), null where no ticker is on record.

Query parameters

integer
Number of articles to return. Between 1 and 500. Defaults to 200.
integer
Number of articles to skip for pagination. Minimum 0. Defaults to 0.
string
Filter the aggregated feed by category. One of all, crypto, tradfi, macro, geopolitics, research, ai. Defaults to returning all categories.
string
Filter articles mentioning a specific coin. Accepts a ticker (BTC, ETH, SOL) or a canonical asset id (bitcoin, ethereum, solana) and resolves it against the canonical coin_tags ids stored with each article.
string
ISO 8601 date string. Only return articles published on or after this timestamp.
string
Search article titles across the stored corpus. Between 1 and 200 characters.
Response

Pagination

Use offset and limit for pagination. Deduplication happens before the offset is applied, so adjacent pages do not overlap when the same story appears from multiple publishers. Use the returned next_offset; a null value means the scan is complete. Some publishers (Reuters, AP News, Messari, The Information and a few others) are read through Google News. Their title is the publisher’s headline, but their link is a news.google.com redirect to the article rather than the publisher’s own URL. total is exact when the bounded corpus scan reaches the end and otherwise is a lower-bound estimate. truncated: true means the safety cap was reached before the complete filtered corpus could be scanned. GET /v1/news/feed?offset=200&limit=200
Read the error reference