MCP App Store
by KeenEthics
Finbar icon

Finbar MCP Server for Claude and ChatGPT

By Finbar

Overview

High-quality financial data and insights on thousands of public companies, designed for AI-native investor workflows. Pull detailed company information, fundamental time series, and market data across a worldwide coverage, backed by a machine-accessible database of millions of company reports, presentations, and transcripts. Trusted many of the world's largest hedge funds and investors.

How to connect

  1. Claude: Open the Finbar listing in the Claude connectors directory and choose Connect. Claude listing Finbar exposes 11 tools here.
  2. ChatGPT: Open the Finbar listing in the ChatGPT app directory and choose Connect. ChatGPT listing Finbar exposes 14 tools here.

Tools

build_dataset

ChatGPT
Populate caller-supplied rows for explicit historical periods. Provide dataset.entity_id, rows with stable id/label/unit/scale, observations (empty for a new dataset), and target_periods as document periods {year, period_type, period}: Y/0 annual, Q/1–4 quarterly, H/1–2 half-year, QC/1–4 cumulative, M/1–12 fiscal monthly, YT/0 or QT/1–4 transitional. Balances use period-end; flows use the reporting interval. No Excel addresses, formulas, row creation or forecasts. May use web search for source discovery; dataset values require Finbar evidence. This starts paid AI work and may take minutes. Actual usage consumes account credits even on cancellation/failure. Reuse idempotency_key for retries. Poll get_dataset_run at its poll_after_seconds interval, then page get_dataset_result; save results within seven days of completion.

cancel_dataset_run

ChatGPT
Request cancellation of your dataset run. Completed observations remain retrievable; incurred AI usage is still charged. Poll until execution and billing finish; billing_state is independent of execution status.

get_account_permissions

ChatGPT
Discover effective account document read allowances, including customer overrides. Free and personal plans allow 10 chunks per request; Enterprise allows 100. All plans normally cap each chunk at 10000 Unicode characters, excluding the abbreviation notice; direct item reads can explicitly disable truncation. Requires documents:read; allowances do not grant OAuth scopes. Call before requesting a larger document window.

get_dataset

ChatGPT
Get metadata for the active published company dataset. Use get_dataset_catalogue to discover queryable historical series.

get_dataset_catalogue

ChatGPT
Discover historical metrics and exact scope paths grouped by reporting basis, with a shared top-level periods array covering the entire dataset, identical on every page. Individual series may have gaps; missing observations return null. Follow next_cursor to read all series. Pass expected_dataset_id to detect a changed publication.

get_dataset_result

ChatGPT
Read a bounded page of dataset observations, corresponding row definitions, changes and diagnostics. limit is an upper bound: default 25, maximum 100. Pages may contain fewer observations to fit 48 KiB; follow next_cursor. Optionally filter row_id and exact period. Follow next_cursor with unchanged filters; assemble pages by row_id and period, not array position. snapshot_status distinguishes interim and final results. Pending targets are still being processed; do not retry them while the run is active. expires_at is this snapshot's expiry; the final snapshot expires seven days after termination. Retrieval uses no AI credits.

get_dataset_run

ChatGPT
Poll dataset execution, current activity (research, verification, correcting or finalizing), current_period for Update, usage and next actions. completed_observations counts saved targets including unresolved entries; reaching the total is not completion. Wait for terminal execution status, then fetch results without a cursor for the final snapshot. Billing is separate: open and pending_settlement may alternate during execution. Wait poll_after_seconds between requests.

get_entity

ChatGPT
Retrieve one Finbar entity by entity_id.

get_entity_wiki

ChatGPT
Read company wiki: OVERVIEW, PRODUCTS, GEOGRAPHY, SEGMENTS, and HISTORY. Optionally filter by types; omitted or empty types returns all available sections. Inline source citations are preserved; use read_document with documents:read to fetch cited document content.

get_market_data

ChatGPT
Retrieve price history and market metadata for a Finbar entity.

list_documents

ChatGPT
List documents for a Finbar entity with optional type, inclusive period, and tag filters. Results use opaque cursor pagination ordered by report_date descending. The tag field is included when non-empty. A non-empty tag is human-readable, free-form metadata that distinguishes documents with the same type and period. It usually identifies a file format, an indexed part split from a master document, or a subsidiary. In a subsidiary document set, an empty tag normally represents the parent consolidated group statements.

query_dataset

ChatGPT
Query historical company dataset series using metric IDs and exact ordered scope combinations from get_dataset_catalogue. Period mode is exact (coordinates), fiscal_range (from/to fiscal ends with zero duration), or trailing (months), bounded to 120. Defaults to original values; latest_restated selects the final populated restatement with original fallback. Returns aligned values and source references. Values retain stored scale; no aggregation. Fetching sources requires documents:read.

read_document

ChatGPT
Read a paragraph/table item or a bounded document window. Document windows require limit and exactly one of start_index or an opaque cursor, and return next_cursor only when another renderable chunk exists.

search_document

ChatGPT
Search paragraph and table content within exactly one document or an entity's documents, optionally filtered by inclusive period bounds. Results use opaque cursor pagination ordered by text score with deterministic paragraph/table tie-breaking.

search_entity

ChatGPT
Search for Finbar entities by name, ticker, or related query. Results include real relevance scores and use opaque cursor pagination ordered by score descending and entity ID ascending.

search_entity_wiki

ChatGPT
Discover companies by searching their dedicated wiki. Optionally filter section types. Returns companies once, with scores and short section previews. Follow next_cursor with the same query and filters within ten minutes; limit may change (1–25, default 10). At most 200 companies will be available in pagination. Use get_entity_wiki for complete text and citations.

update_dataset

ChatGPT
Refresh explicit historical target_periods using an existing structured dataset and its prior observations as matching context. Preserve row IDs and units. Returns proposed changes and a resulting dataset; unresolved refreshes retain prior values with warnings. This starts paid AI work and may take minutes. Actual usage consumes account credits even on cancellation/failure. Reuse idempotency_key for retries. Poll get_dataset_run at its poll_after_seconds interval, then page get_dataset_result; save results within seven days of completion.

get_dataset

Claude

get_dataset_catalogue

Claude

get_entity

Claude

get_entity_wiki

Claude

get_market_data

Claude

list_documents

Claude

query_dataset

Claude

read_document

Claude

search_document

Claude

search_entity

Claude

search_entity_wiki

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/finbar"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/finbar)