# Financial Evidence [Verified owner] [Health: Active]

**Category:** 💰 Finance & Fintech  
**Repository:** https://github.com/beepboop2025/financial-evidence-skills  
**GitHub Stars:** 1  
**Views:** 6  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/financial-evidence

## Description
Read dated financial evidence from an AI agent. Financial Evidence Python 0.1.13 provides three read-only MCP tools: financial_evidence_topics discovers coverage, financial_evidence_route selects sources, and financial_evidence_fetch retrieves public evidence across LiquiLens, Seiche, Undertow and Palimpsest. Review institution disclosures, USD funding conditions and BTC exit estimates with source clocks, citations and coverage limits. Install the verified PyPI release locally, or connect to the separate hosted three-tool router (0.1.8) without an API key. The nine-tool Research Desk is a separate service. This server does not place trades. Setup and first tasks: https://beepboop2025.github.io/financial-evidence-skills/connect/

## Tools
Capabilities this server exposes over MCP:

- **financial_evidence_topics** — List supported topics and their fixed public raw-JSON routes without network access.
- **financial_evidence_route** — Resolve one or more financial research topics to fixed public raw-JSON sources without fetching them.
- **financial_evidence_fetch** — Fetch bounded untrusted read-only JSON from fixed LiquiLens, Undertow, Seiche, and Palimpsest routes. Status reports transport only; evidence is not evaluated and Evidence Carrier verification is not performed.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What Financial Evidence does

Financial Evidence exposes a narrow, read-only research interface over MCP. It helps an agent identify available financial topics, map a selected topic to one or more fixed public sources, and retrieve the corresponding raw JSON. Supported research includes institution disclosures, USD funding conditions, BTC exit estimates, and additional routed topics such as GIFT City, foreign exchange, and gold.

The service is designed for reviews where source dates, citations, and coverage boundaries should remain visible. It can support a funding review, an institution-monitoring task, or a retained funding-and-exit research record. It is not a trading system and does not grant authority to execute trades.

## How it works

The workflow is intentionally split into discovery, routing, and retrieval:

1. `financial_evidence_topics` lists supported topics and their fixed public raw-JSON routes without making a network request.
2. `financial_evidence_route` maps one or more selected research topics to fixed public sources without fetching their contents.
3. `financial_evidence_fetch` retrieves bounded, untrusted, read-only JSON from the available LiquiLens, Undertow, Seiche, and Palimpsest routes.

This separation lets an agent inspect the available coverage and selected sources before requesting evidence. Returned source dates and citations can be retained as part of a review. Transport status reports whether retrieval completed; it does not judge the evidence or perform Evidence Carrier verification.

## Setup and configuration

The hosted deployment uses Streamable HTTP and provides public read-only tools without a local server or API key. It can be connected through the central agent platform or through client-specific MCP configuration for supported clients such as Claude Code, Codex, Cursor, and VS Code.

A verified Python package is also available for local installation. The package supports local product-client workflows, while the separately deployed hosted router maintains its own three-tool interface. Setup instructions cover browser, REST, Python, and remote MCP connection paths. Client or model costs and fair-use limits may apply to connected research workflows.

## Tools and capabilities

- Discover supported topics and fixed public routes without network access.
- Resolve one or more topics to the sources intended for retrieval.
- Fetch bounded JSON from public LiquiLens, Seiche, Undertow, and Palimpsest endpoints.
- Preserve source dates, citations, and stated coverage gaps for later review.
- Research institution disclosures, USD funding conditions, and BTC exit estimates.
- Use public hosted reads without an account or API key.

## Limitations and notes

Fetched data is untrusted raw JSON. The server reports transport status only; it does not evaluate the financial evidence, verify it through Evidence Carrier, or establish that a source is complete or current. Coverage limits and unavailable evidence should therefore remain explicit in downstream reviews.

The tools are read-only and cannot place trades. The public hosted router and local Python package are separate deployment paths, and the hosted router has its own release lifecycle. Research calls may be subject to client or model costs and fair-use limits.

_Full upstream README: https://allmcps.com/mcp/financial-evidence/readme_