Your daily research desk

Bring the evidence into the tools you already use.

Review bank disclosures with LiquiLens, dollar funding with Seiche, and market liquidity with Undertow. Keep the date, source and coverage beside each result.

Try in your browserConnect OpenBBDownload the research kitConnect your agent

No financial-data API key is needed for these public research endpoints. Your analysis platform may require its own account.

Start with a useful question

Three tasks. One reusable setup.

LiquiLens · Bank evidence

Find the latest covered disclosures for ESAF. Review reported asset-quality measures alongside their filing dates and original documents.

bank_risk · Entity ESAF

Explore covered-bank workflows

Seiche · Dollar funding

Review USD benchmarks and their observation dates. Open the dated funding review when you need the documented nine-input reconciliation.

money_markets · Entity USD

Open the funding review

Undertow · Market liquidity

Inspect the available liquidity observations and withheld values. Use the native exit workflow for a named market and trade size.

market_liquidity

Open the exit workflow

These are separate research views. They do not supply credit ratings, executable prices or trade approvals.

OpenBB Workspace

Add the desk once. Return to it for each review.

  1. In OpenBB Workspace, open Add data and add a custom backend.
  2. Use this backend URL:
https://api.seiche.info/openbb

Select Financial Evidence Research Desk under My Apps. Use the Bank Research, Funding & Capital, Market Liquidity and Source Audit tabs. Save your preferred layout in your own workspace.

API and dataset documentation · Python extension setup for OpenBB V4 and V5

This is a public custom backend. Inclusion in OpenBB's curated directory is a separate review.

Python · Jupyter · SQL

Save a cited capture you can use again.

Download research_client.py. Python 3.10 or newer is sufficient; the client has no package dependencies. It follows pagination, checks that source documents remain the same and saves JSON, CSV and a checksum receipt.

python3 research_client.py desk --output captures/first-review
python3 research_client.py bank_risk --entity ESAF --output captures/esaf-review

The first command creates separate funding, bank and liquidity tables. For notebooks, keep the client beside research_desk.ipynb. Open the notebook and run the task you need.

from research_client import collect
funding = collect("money_markets", entity="USD")
rows = funding["results"]

Use each saved rows.csv in R, DuckDB or your firm's approved data tools. Keep evidence.json and receipt.json alongside it.

SELECT entity_id, metric, value, unit, as_of, availability, source_url
FROM read_csv('captures/first-review/money_markets/rows.csv', all_varchar=true);

Captures contain currently published evidence. A historical observation date does not establish that today's dataset was available at that time. Empty and withheld values stay missing.

Excel · Power BI

Refresh a table with its sources attached.

  1. Download FinancialEvidenceRows.pq.
  2. Create a blank Power Query, paste the function into Advanced Editor and name it FinancialEvidenceRows.
  3. Choose Anonymous authentication for https://api.seiche.info.
  4. Invoke a task below and load the table. Refresh when you need a new observation.
FinancialEvidenceRows("money_markets", "USD")
FinancialEvidenceRows("bank_risk", "ESAF")
FinancialEvidenceRows("market_liquidity")

The function returns up to 2,000 rows and refuses to silently truncate a larger result. For larger histories or restricted network environments, import the Python client's saved CSV instead.

MCP · Research agents

Give the agent a clear research starting point.

Add a Streamable HTTP MCP connection with this URL:

https://api.seiche.info/openbb/mcp

Download example configuration and project instructions. Start with financial_evidence_datasets, then financial_evidence_agent_review. On a later review, pass the previous revision to identify changed published evidence.

Discover coverage, then review USD funding, the covered bank ESAF, and market liquidity. Cite the source, date and unit beside every observation. List gaps and the further evidence the analyst needs.

Native LangGraph, CrewAI, OpenAI Agents and Pydantic AI integrations · n8n and scheduled research examples

Bloomberg · Other institutional platforms

Bring a reviewable data capture to your institutional workspace.

The research kit supplies portable Python, CSV and JSON for your firm's approved import process. A Bloomberg BQuant administrator can assess these assets for an entitled environment; our notebook has not been validated inside BQuant or approved for Bloomberg App Portal distribution.

For LSEG Workspace, FactSet and other vendor platforms, use the firm's supported external-data process or the FDC3 reference app where supported. Platform acceptance and entitlements remain separate from portable file compatibility.

Technical evaluation and vendor submission materials · Setup, limits and repeat-use checklist