io.github.cyanheads/fdic-banks-mcp-server

Compare one institution with a peer group for one quarter: for each metric, the institution's value next to the peer median, quartiles, minimum and maximum, and its percentile and rank. The default peer group is every institution in the same asset-size band that reported that quarter; narrow it to one state, widen it to all sizes, or name the peers by CERT. Dollar amounts are in thousands; ratios are percentages.

Describe the df_<id> dataframes that fdic_query_financials and fdic_get_deposits staged when a result at offset 0 was larger than one page, plus any that fdic_dataframe_query saved with register_as. Pass name (from a dataset field) for one dataframe in full: source tool, the parameters it was called with, row count, creation and expiry time, column schema, and the unit and basis of each amount, ratio, or count column (thousands of dollars, percent, or count). Read a table's columns this way before writing SQL for fdic_dataframe_query. Omit name to list the live dataframes, newest first, 50 per page, as name, source tool, row count, and expiry only. A deployment that serves HTTP without authentication, where every caller shares one canvas, does not list; there, pass the name a dataset field returned.

Run one read-only SELECT (DuckDB SQL) across the df_<id> dataframes that fdic_query_financials and fdic_get_deposits staged or an earlier register_as saved; joins, aggregates, window functions, and CTEs work. Before writing SQL, pass each table name to fdic_dataframe_describe to read its columns. DOUBLE columns come back as JSON numbers and dollar columns are thousands of US dollars; BIGINT results such as COUNT(*) come back as strings, so CAST them to INTEGER or DOUBLE for arithmetic. Writes, DDL, file-reading functions, and system catalogs are rejected. register_as materializes the result as a new dataframe with a fresh TTL.

Get Summary of Deposits data (branch-level domestic deposits, annual as of June 30, 1994 onward). With cert only: the institution's branches and its deposit market share in each state where it has offices. With a geography (state, county, city, ZIP, or MSA code): every institution in that market ranked by deposits, with market share and the Herfindahl-Hirschman index. With both: the institution's branches in that market and its rank and share there. Branches and rankings come a page at a time, continued with offset; at offset 0 a set larger than the page is also staged as a dataframe for fdic_dataframe_query. Defaults to the latest survey year. Dollar amounts are in thousands.

Get one institution's quarterly Call Report financials by CERT — balance sheet, income, returns, credit quality, and capital ratios — most recent quarter first, with its name, status, and holding company. Unsuffixed income and return metrics are single-quarter figures; _ytd metrics accumulate from January 1. Dollar amounts are in thousands. Quarterly data lands about seven weeks after quarter end; history reaches back to 1984.

Get one institution's structure-change history by CERT, newest first: the mergers it absorbed or was absorbed in, failures and assisted takeovers, branch purchases and sales, renames, relocations, charter and regulator changes, openings, and closings, each with its effective date, FDIC change code, the institution's role, and the other institution's CERT and name. A merger or failure appears once, from whichever side the CERT was on. Branch-level openings, closings, sales, and relocations are left out unless event_types includes branch. Merger records start in 1969 and failure records in 1970.

List the vocabulary the other fdic_ tools accept and return: every financial metric name with its FDIC field, unit, and basis (single quarter, year-to-date, or point in time), bank charter classes, failure resolution methods, the structure-change codes fdic_get_institution_history reports and the event type each maps to, insurance funds, the asset bands fdic_compare_peers uses, and the years each dataset covers. Served from built-in tables; no request to FDIC.

Pull a multi-institution, multi-quarter Call Report panel — one row per institution per quarter — filtered by CERTs, headquarters state, asset range, and thresholds on any catalog metric. Use it to screen (every bank in a state with a noncurrent-loan rate above 3%) or to build a trend panel for SQL. Returns one page of the panel sorted as requested, continued with offset; at offset 0 a panel larger than the page is also staged as a dataframe for fdic_dataframe_query. With no dates it covers the latest published quarter only.

Search FDIC-insured bank failures and assistance transactions since 1934 by name, CERT, headquarters state, failure date range, resolution method, or size. Returns each event with failure date, acquirer, total assets and deposits, and the FDIC's estimated loss to the insurance fund, plus totals over every matching event; group_by adds counts and losses per year, state, method, or fund. Searches failures only unless resolution is set to assistance or all. Dollar amounts are in thousands.

Find FDIC-insured banks and savings institutions by name, CERT, location, size, charter class, or holding company — including closed, merged, and failed institutions. Name matching is fuzzy, case-insensitive, and also matches former and trade names (a result says which it matched); every word must match. Returns institution records keyed by CERT, the FDIC certificate number every other fdic_ tool takes, with active status, successor CERT for merged or failed institutions, holding company, and latest reported assets and deposits. Credit unions are insured by the NCUA and are not in this data.