Tools

The MCP server exposes 13 read-only tools across two layers:

Event-level: clustered and analyzed. Carries divergence, market relevance, market sentiment, fact ledger, per-outlet framing.
Article-level: individual articles tagged with ~720 ticker symbols and 47 topics. Pre-clustering, so duplicates from different outlets are expected. Broader ticker coverage than event-level.

All tools return JSON strings. Empty results return {"results": [], "count": 0, "message": "..."} rather than an error.

get_digest

Zero-parameter morning briefing. The right starting point for any session.

get_digest()

Returns a 24-hour summary: top market events, sentiment breakdown, most contested stories, category counts, and top tickers. No parameters needed.

Response fields

Field	Type	Description
`generated_at`	ISO 8601	When the digest was built
`window_hours`	int	Always 24
`market_summary.total_events`	int	All events in the window
`market_summary.market_events`	int	Events with market_relevance_score ≥ 70
`market_summary.sentiment`	object	`{bullish, bearish, neutral}` counts
`market_summary.top_tickers`	list	Most-mentioned tickers, up to 10
`top_market_events`	list	Up to 5 events sorted by market relevance
`most_contested`	list	Up to 5 events sorted by divergence score
`categories`	object	Event count by category

get_market_news

Recent events with material market relevance.

get_market_news(limit=10, hours=24, min_score=70)

Parameters

Name	Type	Default	Notes
`limit`	int	10	1 to 25
`hours`	int	24	Lookback window. Max 168 (7 days)
`min_score`	int	70	Minimum market_relevance_score. Use 50 for broader results, 85+ for high-impact only

get_market_articles

Article-level market desk items. Pre-clustering, so duplicates from different outlets covering the same story are expected.

get_market_articles(tag="", tags="", outlet="", hours=168, limit=25, offset=0)

Two filter modes:

tag - single exact tag, e.g. "ticker:NVDA" or "topic:fed".
tags - comma-separated tags, all must match (intersection), e.g. "ticker:LLY,topic:fda" for Eli Lilly FDA articles only.

Tag taxonomy

ticker:<SYMBOL> - ~720 names: equities and ETFs. Use get_market_tags(tag_type="ticker") to discover what's active.

topic:<NAME> - 47 topics:

Group	Topics
Central banks	`fed`, `ecb`, `boe`, `boj`
Macro / data	`rates`, `inflation`, `cpi`, `ppi`, `gdp`, `jobs`, `recession`, `confidence`, `retail`
Markets / instruments	`indices`, `treasuries`, `bonds`, `credit`, `pe_vc`, `fx`
Corporate	`earnings`, `guidance`, `ipo`, `m&a`, `layoffs`, `dividend`, `bankruptcy`
Trade / policy	`trade`, `sanctions`, `antitrust`, `regulation`, `litigation`
Healthcare	`fda`
Management	`management`, `labor`
Tech / cyber	`ai`, `cyber`
Multilateral	`imf`, `rating`
Corporate events	`contract`, `split`, `offering`
Crypto / commodities	`crypto`, `oil`, `gold`, `commodities`
Catch-all	`markets`, `analyst`

Parameters

Name	Type	Default	Notes
`tag`	string	-	Single exact tag. Ignored if `tags` is provided
`tags`	string	-	Comma-separated AND-filter
`outlet`	string	-	Exact outlet name
`hours`	int	168	Lookback. Max 336 (14 days)
`limit`	int	25	1 to 50
`offset`	int	0	Pagination offset

Response fields

Field	Description
`results`	Article rows with `id`, `headline`, `outlet`, `url`, `image_url`, `published_at`, `fetched_at`, `sentiment`, `bias_label`, `tickers`, `topics`, `tags`
`count`	Number of rows returned
`tag`	Single-tag filter used (if any)
`tags`	Multi-tag filter used (if any)
`outlet`, `offset`, `limit`	Echo of inputs

get_market_tags

Most common market classifier tags in recent article-level data.

get_market_tags(hours=168, limit=50, tag_type="all")

Use this before get_market_articles to discover active ticker and topic tags, or to build a current "hot tickers" or "hot topics" scan.

Parameters

Name	Type	Default	Notes
`hours`	int	168	Lookback. Max 336 (14 days)
`limit`	int	50	1 to 200
`tag_type`	enum	`all`	`ticker`, `topic`, or `all`

Response fields

Field	Description
`results`	Tag rows with `tag` and `count`, sorted by count descending
`count`	Number of tag rows returned
`window_hours`	Echo of input
`tag_type`	Echo of input

get_ticker_news

Event-level lookup for a specific ticker (last 30 days).

get_ticker_news(ticker)

Returns clustered events with divergence and sentiment for a given ticker. If empty, try get_ticker_articles for article-level coverage, which has broader ticker recognition including smaller-cap and foreign issuers.

Parameters

Name	Type	Notes
`ticker`	string	Required. Case-insensitive. No `$` prefix.

get_ticker_articles

Article-level lookup for a specific ticker. Wider coverage than get_ticker_news: ~720 names including foreign issuers (SFTBY, BAYRY, NU), small caps (TEM, RXRX), and ETFs (SPY, XBI, IBIT).

get_ticker_articles(ticker, hours=168, limit=25)

Parameters

Name	Type	Default	Notes
`ticker`	string	required	Case-insensitive. No `$` prefix
`hours`	int	168	Lookback. Max 336 (14 days)
`limit`	int	25	1 to 50

get_trending_tickers

Surfaces tickers with disproportionate recent mention activity compared to a baseline window. Unlike get_market_tags (ranked by volume), this ranks by acceleration.

Returns ratio = recent_per_hour / baseline_per_hour. Ratios above 3 indicate a story is breaking around the name; above 6 indicates a major event.

get_trending_tickers(hours=24, baseline_hours=168, limit=10, min_recent_mentions=2)

Parameters

Name	Type	Default	Notes
`hours`	int	24	Recent window
`baseline_hours`	int	168	Baseline window. Must be larger than `hours`
`limit`	int	10	1 to 25
`min_recent_mentions`	int	2	Excludes one-off mentions

Response fields per row

Field	Description
`ticker`	Symbol
`recent_mentions`	Mention count in `hours` window
`baseline_mentions`	Mention count in `baseline_hours` window
`ratio`	recent rate / baseline rate, rounded to 2 decimals

scan_portfolio

Batch ticker check. Returns hits, no-news tickers, and an aggregate sentiment breakdown.

scan_portfolio(tickers, hours=24, include_articles=False)

Parameters

Name	Type	Default	Notes
`tickers`	string	required	Comma-separated. E.g. `"NVDA,AAPL,MSFT"`. Max 20, excess silently dropped
`hours`	int	24	Lookback. Max 168 (events) / 336 (articles)
`include_articles`	bool	false	Also surface per-ticker article-level hits, which cover foreign and smaller-cap names that event-level may miss

Response fields

Field	Description
`hits`	Tickers with at least one event, each with a list of matching events
`no_news`	Tickers with no events (and no articles, when `include_articles=true`) in the window
`tickers_checked`	Echo of inputs (capped)
`overall_sentiment`	`{bullish, bearish, neutral}` counts across all hits
`article_hits`	Present when `include_articles=true`. Per-ticker article-level matches (up to 10 each)

get_market_sentiment_overview

Aggregate bullish/bearish breakdown for the market as a whole.

get_market_sentiment_overview(hours=48, min_score=70)

Parameters

Name	Type	Default	Notes
`hours`	int	48	Lookback. Max 168
`min_score`	int	70	Minimum market_relevance_score to include

Response fields

Field	Description
`window_hours`	Echo of input
`total_market_events`	Events meeting the score threshold
`sentiment_breakdown`	`{bullish, bearish, neutral}` counts
`top_tickers`	Most-mentioned tickers in market events
`most_divergent`	Up to 5 highest-divergence market events

query_news

Power query. Combine any filters in a single call.

query_news(tickers="", sentiment="", category="", min_divergence=0,
           max_divergence=100, min_market_score=0, hours=24,
           sort="recent", limit=10)

Parameters

Name	Type	Default	Notes
`tickers`	string	-	Comma-separated ticker symbols
`sentiment`	enum	-	`bullish`, `bearish`, `neutral`
`category`	enum	-	`war`, `economy`, `politics`, `ai`, `health`, `other`
`min_divergence`	int	0	0 to 100
`max_divergence`	int	100	0 to 100
`min_market_score`	int	0	Minimum market_relevance_score
`hours`	int	24	Lookback. Max 168
`sort`	enum	`recent`	`recent`, `divergence`, `market`
`limit`	int	10	1 to 25

get_top_divergent

Most editorially contested stories, sorted by divergence score.

get_top_divergent(category="", hours=24, limit=10)

Returns events sorted by divergence_score descending. Each result includes a takeaway sentence summarising the editorial split. Use get_event for the full per-outlet breakdown.

Parameters

Name	Type	Default	Notes
`category`	enum	-	`war`, `economy`, `politics`, `ai`, `health`, `other`. Empty = all
`hours`	int	24	Lookback. Max 168
`limit`	int	10	1 to 25

search_news

Keyword search across event titles and summaries.

search_news(query, category="", sentiment="", min_divergence=0, limit=10)

Case-insensitive substring match. Does not search raw article text.

Parameters

Name	Type	Default	Notes
`query`	string	required	Matched against title and event_summary
`category`	enum	-	`war`, `economy`, `politics`, `ai`, `health`, `other`
`sentiment`	enum	-	`bullish`, `bearish`, `neutral`
`min_divergence`	int	0	0 to 100
`limit`	int	10	1 to 25

get_event

Full event detail: sources, facts, takeaway, and article count.

get_event(event_id)

Parameters

Name	Type	Notes
`event_id`	int	From the `id` field on any other tool result

Response fields

Field	Description
`id`	Event ID
`title`	Event title
`summary`	2–3 sentence factual summary
`takeaway`	Summary of how outlet coverage diverges on this event
`category`	`war`, `economy`, `politics`, `ai`, `health`, `other`
`divergence_score`	0–100. See Divergence score
`market_relevance_score`	0–100, or null
`market_sentiment`	`bullish`, `bearish`, `neutral`, or null
`market_relevance_reason`	Why this event is market-relevant, or null
`tickers`	Stock ticker symbols extracted from coverage
`companies`	Company names mentioned
`sources`	Per-outlet: `outlet`, `headline`, `framing_label`, `framing_color`
`facts`	Per-fact: `status` (confirmed/disputed), `text`, `source_attribution`
`article_count`	Articles clustered into this event

Returns {"error": "not_found"} if the event ID does not exist or the event was rejected during analysis.