Name: Forecast Accuracy & Track Record
Author: GeminiGeorge22

Forecast Accuracy & Track Record MCP

Forecast track records: forecasts vs actuals, earnings beats, prediction markets, archived news.

Grade A Remote or local · Streamable HTTP · stdio API key 7 tools v0.2.1 · 3 Nov 2026 MIT

No reviews yet by GeminiGeorge22PyPI 765 Runs on your machine or at its publisher’s address — not through us

100/100

trust score

downloads in 30d

local

or remote

tools declared

What it does

Forecast track records: forecasts vs actuals, earnings beats, prediction markets, archived news.

Quickstart

# 1 — add it to Claude Code
claude mcp add --scope user forecast-accuracy-track-record-mcp '--' uvx forecast-accuracy-mcp

# 2 — ask your agent for something these tools do
#   publicrecords--calibration-ledger, pintostudio--economic-calendar-data-investing-com, scrapemint--stock-earnings-estimates and 4 more

Forecast Accuracy & Track Record costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.

Publishernot claimed

GeminiGeorge2211 servers

6 categories

All 11 servers

Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.

Installation

Where are you running it?

Every route below installs the same thing and ends at the same approval screen. This one can run either way — at an address its publisher runs, or as a process under your own user, with your files and your network — and this marketplace is in the path of neither.

Claude Codedeclared by this listing

This is a public server: you run it yourself and this marketplace is not in the path. You need uv (docs.astral.sh/uv): uvx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. The '--' is in quotes so that PowerShell passes it on.

claude mcp add --scope user forecast-accuracy-track-record-mcp '--' uvx forecast-accuracy-mcp

Reconnect, or start a new session, and the tools appear in the model’s tool list.

Or connect to its publisher’s address

Its publisher also runs it at https://mcp.apify.com/?tools=publicrecords/calibration-ledger,pintostudio/economic-calendar-data-investing-com,scrapemint/stock-earnings-estimates,fatihtahta/kalshi-scraper,parsebird/polymarket-scraper,ryanclinton/wayback-machine-search,data_xplorer/google-news-scraper-fast: nothing to start on your machine, and this marketplace is not in the path.

claude mcp add --scope user --transport http forecast-accuracy-track-record-mcp 'https://mcp.apify.com/?tools=publicrecords/calibration-ledger,pintostudio/economic-calendar-data-investing-com,scrapemint/stock-earnings-estimates,fatihtahta/kalshi-scraper,parsebird/polymarket-scraper,ryanclinton/wayback-machine-search,data_xplorer/google-news-scraper-fast'

Free on mcprush, and it runs from its publisher’s own package: this marketplace is not in the path, so there is no key of ours to hold and no card to add.

What it asks of you

It reaches a system of yours, so it needs your own credential rather than the publisher’s: Authorization header — The API key mcp.apify.com issues — sent as Authorization: Bearer <key>. You set it in your own client’s config, on your machine — this marketplace never holds it.

What it writes

One config entry your client uses to start the process locally. A local server runs with your file system and your network, which is why it is priced without metering.

What you approve

7 tools, with what each one reads, writes and reaches shown before you agree — the same list on every route above. Read the tool surface.

Tool surface

What the model actually sees. Descriptions are diffed on every release — see version history.

Only tools that write or reach the network

Read from its source code. A server that builds its tool list when it starts can list more tools than this.

Calibration Ledger — trust a forecast, with evidence (beta). Given a claim from an organization (IMF, OECD, Fed/FOMC, SPF, ECB SPF, or a listed issuer), returns that organization's track record: error distribution, intervals, bias, band coverage, worst misses with source links — or honest 'no…

Parameter	Type	Required	Notes
mode	string	yes	text = hand it a sentence ("The IMF expects Canada's economy to grow 1.8% in 2027"). lookup/score/compare use structured fields.
claim_text	string	no	Free-text claim for text mode (e.g. "The IMF expects Canada's economy to grow 1.8% in 2027"). Example: "The IMF expects Canada's economy to grow 1.8% in 2027."
source	string	no	imf_weo \| oecd_eo \| fed_sep \| spf \| ecb_spf \| pbo, or an issuer name; optional in text mode Example: "oecd_eo"
series	string	no	Generated from rules.yaml — every series the ledger resolves. Example: "RGDP_GROWTH_A"
geo	string	no	ISO3 country code, EA, or USA. Required for most score/lookup/compare calls. Example: "CAN"
target_period	string	no	Target period as YYYY, YYYYQn, or YYYY-MM when looking up a specific claim. Example: "2027"
point	string	no	Numeric claimed value for lookup mode (string so decimals stay exact). Example: "1.8"
published_at	string	no	When the claim was published (YYYY-MM-DD), for lookup/verify. Example: "2026-06-01"
horizon_bucket	string	no	Optional horizon bucket filter for score mode.
band_kind	string	no	For Fed SEP score mode: central_tendency or range.
claim_id	string	no	Ledger claim id for verify mode.
issuer	string	no	Organization name as it appears on its releases/filings (e.g. Salesforce Inc). Both records are returned: dated promises (newswires) and numeric guidance (EDGAR).

Economic Calendar Data (Investing.com). Extract structured economic calendar events (actual, forecast, previous, importance) from Investing.com across 45+ countries. Built for algorithmic trading systems, AI trading agents, RAG pipelines, and financial research. Filter by country, importance,…

Parameter	Type	Required	Notes
timeZone	string	no	Time zone in GMT +/- hours:minutes format (e.g., 'GMT +2:00'). If not provided, the local GMT time zone will be used.
timeFilter	string	no	Determines how time is displayed: 'time_only' shows the event time, 'time_remain' shows remaining time until event. Default is 'time_only'. Options: time_only = Time Only; time_remain = Time Remaining
importances	string	no	List of importance ratings for events. Can contain 'high', 'medium', and 'low'. If empty or null, all importance ratings including holidays will be considered. Options: high = High; medium = Medium; l
categories	string	no	List of categories to which the events will be related to. If empty or null, all available categories will be taken into consideration. Example: ""
country	string	no	A single country to retrieve economic calendar events for (e.g. United States, Euro Zone). To select several countries, use "Countries (multiple)" below. Leave both blank to include all countries. Exa
countries	array	no	Select one or more countries/regions. Combined with "Country" above. Leave empty (and Country blank) to include all countries.
fromDate	string	no	Start date for retrieving the economic calendar in 'YYYY-MM-DD' format. If not provided, only the current day's economic calendar will be retrieved.
toDate	string	no	End date for retrieving the economic calendar in 'YYYY-MM-DD' format. If not provided, only the current day's economic calendar will be retrieved.

Stock Earnings Estimates & Results: EPS Forecast vs Actual. Keyless earnings data for any US stock from NASDAQ: what analysts expect the company to earn next quarter, what it actually earned in past quarters, the percentage it beat or missed by, how often it has beaten, and whether analysts raised…

Parameter	Type	Required	Notes
mode	string	no	summary = one row per ticker: what analysts expect next quarter, what the company delivered last quarter, how often it has beaten, and whether estimates are being raised or cut. surprises = one row pe
symbols	array	no	US tickers to look up, e.g. NVDA, TSLA, AAPL. Maximum 200 per run.
minBeatRate	integer	no	Keep only companies that beat expectations in at least this share of their reported quarters. 75 returns serial beaters. 0 = no filter.
onlyRaisedEstimates	boolean	no	Keep only tickers where more analysts raised their forecast than cut it in the last four weeks. Estimate revision direction is the forward looking half of this data.
includeAnnual	boolean	no	Return full year estimates alongside the quarterly ones.
maxRows	integer	no	Cap on rows returned across all tickers. Controls total cost.

Kalshi Scraper | Fast & Reliable. Get live prediction market data from Kalshi.com with tickers, categories, prices, and volume metrics. Fast, reliable, and structured for analysis. Ideal for traders, researchers, and data teams. When to use: Use when you need current Kalshi prediction-market…

Parameter	Type	Required	Notes
startUrls	array	no	Paste any Kalshi market or search result URLs that you want the actor to visit.
targetUrl	string	no	Paste one Kalshi search or market URL for a quick run.
targetUrls	array	no	Optional alias for startUrls. Paste more Kalshi search or market URLs here.
searchQueries	array	no	List the keywords you want to search for.
maxResults	integer	no	Limit how many market listings the actor should collect before stopping.
pageSize	integer	no	How many search results to fetch per API call. The actor caps this between 1 and 100.
proxyConfiguration	object	no	By default it uses Apify Residential proxies.

Polymarket Leaderboard Scraper. Scrape Polymarket leaderboard rankings with full trader profiles. Extract wallet addresses, usernames, profit/loss, volume, positions, recent trades, bios, and social links. Filter by time range, category, and ranking type. When to use: Use when you need Polymarket…

Parameter	Type	Required	Notes
timeRange	string	no	Leaderboard time period. DAY = today, WEEK = this week, MONTH = this month, ALL = all time. Options: DAY = Today; WEEK = This Week; MONTH = This Month; ALL = All Time.
category	string	no	Market category to filter the leaderboard. OVERALL includes all categories. Options: OVERALL = All Categories; POLITICS = Politics; SPORTS = Sports; CRYPTO = Crypto; FINANCE = Finance; CULTURE = Cultu
leaderboardType	string	no	Rank traders by profit/loss (PNL) or trading volume (VOL). Options: PNL = Profit / Loss; VOL = Volume.
maxResults	integer	no	Maximum number of traders to scrape from the leaderboard (1-1000).
includePositions	boolean	no	Fetch current open positions for each trader. Adds position details like market title, outcome, size, average price, current value, and PnL.
includeActivity	boolean	no	Fetch the last 10 trades for each trader. Adds trade details like market title, outcome, side (buy/sell), size, price, and timestamp.

Wayback Machine Scraper - Track Website Changes Over Time. Search the Internet Archive's Wayback Machine for historical snapshots of any website. Retrieve archived page metadata -- including timestamps, URLs, MIME types, HTTP status codes, and content hashes -- for up to 10,000 snapshots per run.…

Parameter	Type	Required	Notes
monitor	boolean	no	Turn on for scheduled monitoring runs. One flag enables: change detection, only-changed snapshots, change intelligence + diffs, alerts on moderate-or-major changes, timeline output, insights record, A
deltaSinceLastRun	boolean	no	Turn on for stateful monitoring without external storage. The actor remembers the events and snapshot URLs returned in the previous run (in a named key-value store) and filters this run to only the ev
monitorStateKey	string	no	Optional override for the key-value-store key used by Delta mode. Auto-derived from the sorted URL list if not set; provide a stable string here if you want to share state across separate runs that ha
useCase	string	no	Pick a preset to auto-configure filters, change detection, and output mode for a common workflow. Set to "custom" to use the raw inputs below. SEO = monthly buckets + 200-only + change-only. Complianc
url	string	no	URL or domain to search for snapshots (e.g., "example.com" or "https://example.com/page"). Ignored if you supply a list under URLs (batch). Console UI prefills "apify.com" as a placeholder; programmat
urls	array	no	Optional: query multiple URLs in one run for comparative competitive intelligence. Each URL gets its own snapshots, change detection, and metrics in the output. The single URL above is added if both f
matchType	string	no	How to match the URL. "exact" = exact URL only, "prefix" = URL prefix match, "host" = all pages on same host, "domain" = all subdomains too Options: exact = Exact URL; prefix = URL Prefix; host = Same
dateFrom	string	no	Start date filter (YYYYMMDD or YYYY format, e.g., "20200101" or "2020")
dateTo	string	no	End date filter (YYYYMMDD or YYYY format, e.g., "20231231" or "2023")
targetDate	string	no	Return the single snapshot closest to this date — useful for legal/compliance evidence ("what did the page say on 2020-06-15?"). Accepts YYYY-MM-DD or YYYYMMDD. When set, overrides Max Results and ret
statusFilter	string	no	HTTP status code filter (e.g., "200" for successful pages only)
mimeFilter	string	no	MIME type filter (e.g., "text/html" for HTML pages only)
collapseBy	string	no	Collapse duplicate snapshots. "digest" = unique content only, "timestamp:6" = one per month, "timestamp:8" = one per day, "timestamp:10" = one per hour Options: = No collapsing; digest = Unique conte
fastLatest	boolean	no	Return the most recent snapshots faster by skipping older index segments. Only useful for domains with hundreds of thousands of snapshots — combine with Max Results to fetch the latest N quickly.
autoPaginate	boolean	no	When the CDX API hits its 10,000-row cap, automatically chunk the query year-by-year and merge the results — letting you pull full multi-year domain histories in one run. Costs more proxy time; only e
maxResults	integer	no	Maximum number of snapshots to return per URL (max 10,000 in a single CDX query — enable Auto-paginate to exceed). Ignored when Target Date is set (always returns one snapshot).
detectChanges	boolean	no	Compare each snapshot to the previous one and flag rows where the content hash or HTTP status changed. Adds a "changed" boolean and "changeType" string to every result. Free — uses metadata you alread
onlyChanged	boolean	no	Filter results to only the snapshots where the page actually changed (plus the first "initial" baseline). Requires Detect Changes. Turns 1,000 raw snapshots into ~50 actionable rows.
changeIntelligence	boolean	no	Augment every changed row with a changeSummary (magnitude: minor/moderate/major, categories: pricing/legal/product/layout/etc., keyDiffs: human-readable change phrases) and a structured diff (added/re
alertOnMagnitude	string	no	Emit a single "alert" record at the top of the dataset when changes meet the threshold. Pairs well with scheduling: connect a webhook → Slack/email when the alert record fires. Requires Change Intelli

Google News Scraper. Scrape Google News in real time, including images and descriptions. This tool extracts complete structured data: high-resolution visuals, full, titles, sources, dates, and direct URLs. When to use: Use when you need recent Google News articles for a query (title, source,…

Parameter	Type	Required	Notes
keywords	array	no	List of keywords to search for. You can use Google News search operators: - Use '-' to exclude words (e.g. 'bitcoin -ethereum') - Use 'site:' to filter by specific website (e.g. 'bitcoin site:cnn.com'
topics	array	no	Select one or more predefined Google News topics to scrape.
topicUrls	array	no	Provide custom Google News topic or section URLs to scrape. Useful for niche topics not in the predefined list. Example: https://news.google.com/topics/CAAqJggKIiBDQkFTRWdvSUwyMHZNRGx1YlY4U0FtVnVHZ0pW
maxArticles	integer	no	Maximum number of news to extract per keyword or topic (0 = no limit)
timeframe	string	no	Time period for the news search. Note: this filter applies mainly to keyword searches. Topic pages return their own curated results. Options: 1h = Last hour; 1d = Last day; 7d = Last week; 30d = Last
region_language	string	no	Region and language for news search
decodeUrls	boolean	no	If enabled, decodes Google News URLs to get the original article URLs. Note: This will significantly slow down the scraping process as each URL needs to be decoded individually.
extractDescriptions	boolean	no	If enabled, extracts the meta description from each article page. Note: This will significantly slow down the scraping process.
extractImages	boolean	no	Fetches the og:image from each article page when no image is available (mainly for RSS/topic results). When this option is not explicitly set, images are extracted by default. Toggle OFF to skip image
proxyConfiguration	object	no	Configure Apify Proxy. Residential proxies recommended for best results.

Free

$0 forever

Every tool, no call limit
No card, and mcprush charges nothing for it
Source published under MIT
Runs on your machine — nothing of it reaches our gateway
Nothing to cap, because mcprush bills nothing

Nothing here is billable: mcprush charges nothing to install Forecast Accuracy & Track Record or to call it, at any volume.

ATrust score 100 / 100

Release 0.2.1 · scanned 7 Nov 2026

Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.

Threat score100

Client score100

Findings0 threat · 0 capability

Capabilityminimal

Coveragesource

Verificationsource

Surface digest7598766d170a8867b06e20af8a7d4a22963ceb1a4bccb001a350e885e37ca273

Score vectorthreat 100 → client 100

Term	Level	What it prices	Points
capability-exposure	minimal	capability blast radius (minimal) — client exposure if the model is manipulated	−0
verification-discount	source	publisher verification (provenance) — cryptographic build provenance ties the artifact to its source	−0
coverage-honesty	source	inspection depth (source) — how much of the target the scan could see	−0

score = clamp(round(threatScore − E_cap − E_ver − E_cov), 0, 100). No model, no opinion, no network call at scoring time: the same methodology version against the same bytes always produces this number, and every line above is a field of the report rather than a summary of it.

What the scan could actually read

A grade is only as meaningful as its coverage, so the scanner publishes its own depth before it publishes its result.

Tool surfacenot available

Implementation sourceread

Package metadataread

Live transportnot available

No tools were enumerated, so prompt-injection, capability and toxic-flow analysis had no tool surface to inspect. To grade a package’s real runtime tools, scan the running server: --command "npx -y <package>".

Capability — what it could do if the model were manipulated

Tags derived from each tool’s schema and the implementation, not from what the tool calls itself. minimal is the level these add up to.

Tool	Capability tags	Why the tag was assigned
No per-tool rows for this release — see the note below.

No tool surface was enumerated, so there is nothing to attribute a capability to.

Toxic-flow graph

The lethal trifecta, checked as a graph rather than as a checklist: untrusted input, a sensitive source and an external sink have to meet before there is a path worth worrying about.

The public result for this release does not print the flow graph, so there is nothing to show here. That is not the same as "no paths were found": what the scan did read is above, under coverage.

Supply chain and provenance

Integrityfirst-seen

Surface digest7598766d170a8867

Scanned2026-11-07

This is the first scan of this surface here, so there is nothing yet to compare it against.

Every result on this tab comes from one deterministic pass over the published package — offline, rule by rule, and auditable line by line above. Same methodology version, same bytes, same score.