# EdgeDepth Research [Health: Active]

**Category:** 💰 Finance & Fintech  
**Repository:** https://github.com/edgedepthhq/edgedepth-research-mcp  
**GitHub Stars:** 3  
**npm Downloads (last month):** 1724  
**Views:** 1  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/edgedepth-research-2

## Description
Search recorded crypto and TradFi microstructure through deterministic, reproducible agent tools.

## Tools
Capabilities this server exposes over MCP:

- **list_features** — Returns the closed grammar registry: feature ids, types, ranges, operators, windows, sequence rules, limits, and error codes. `search`, `feature_ids` and `compact` narrow it.
- **list_instruments** — Returns the research universe and coverage. The default is a compact summary; use `symbols: [...]` for selected full records or `full: true` for the verbatim canonical universe.
- **prepare_study** — Free deterministic structured preparation, provenance and allowance estimate. Requires the web `/prepare` release first.
- **interpret_prose** — Turns prose into a proposed query document. It does not execute the query. Optional `time_zone` accepts an IANA time zone for calendar planning.
- **run_scan** — Executes a `research_query.v2` document and returns result bytes with counts, denominators, outcomes, the unconditional same-scope reference, and the reproducibility key. Projected by default (`rows`, `full_rows`, `full_counts`).
- **next_page** — Continues a prior scan with its opaque cursor. Never construct cursors manually.
- **ground_screenshots** — Resolves host-extracted screenshot coordinates against recorded candle closes and coverage, retaining uncertainty and deduplicating event views. Free.
- **investigate_move** — Reads the existing lead-up and optional recorded detector geometry for a grounded event, and optionally prepares exact unrun setup documents. Free read; historical entitlement applies.
- **snapshot_at** — Reads registry feature values, window aggregates, and fired rules as of a recorded moment.
- **base_rate** — Counts matches and eligible buckets for one clause over a window.
- **commonality** — Finds the deterministic intersection across multiple moments with selection-bias caveats included.
- **get_report** — Retrieves a published report by its 8-character canonical hash.
- **run_cohort** — Compares what followed every match with what followed every other eligible predicate-false bucket.
- **run_stratified** — Partitions one matched population at its existing anchors into split-true, split-false, and split-absent outcome summaries.
- **outcome_first** — Starts from the MOVE instead of the setup: names an outcome (size, direction, horizon) and reports what the record was doing at five fixed offsets before every realised move like it. Each row carries two counted shares, the share before these moves and the share across every eligible minute in the…

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "edgedepth-research": {
    "command": "npx",
    "args": ["-y","@edgedepth/research-mcp"],
    "env": {
      "EDGEDEPTH_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `EDGEDEPTH_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What EdgeDepth Research does

The EdgeDepth Research MCP server gives MCP clients access to EdgeDepth's recorded-market research system for Binance USDT-M crypto perpetuals. It is designed for studies that need an exact condition, a defined eligible population, forward outcomes, and an auditable result rather than a single illustrative chart or example.

The server can describe the available feature grammar and research universe, turn a natural-language request into a proposed query document, execute a `research_query.v2` document, and continue paginated scans. It also supports event-focused analysis, including screenshot grounding, move investigation, point-in-time feature snapshots, base rates, commonality, cohort comparisons, and stratified outcomes.

## How it works

EdgeDepth Research MCP server uses one tool core through hosted Streamable HTTP or local stdio. The hosted endpoint is `https://mcp.edgedepth.com/mcp` and uses browser authorization. Local execution runs the npm package with `npx` and an API key.

A typical workflow starts by checking the grammar or interpreting the research question. `prepare_study` can validate a structured scope, predicates, and outcome and return an unrun definition with an allowance estimate. `interpret_prose` produces a proposed query but does not execute it. After approval, `run_scan` executes the unchanged document. `next_page` uses the cursor returned by a previous scan; cursors should not be constructed manually.

Scan-family responses are projected by default to reduce client context usage. Options such as `full_rows`, `full_outcomes`, and `full_counts` recover additional detail where supported. Counts, denominators, exclusions, outcome coverage, and the reproducibility key remain part of the result contract.

## Setup and configuration

For hosted use, add the MCP URL to a compatible client and complete the EdgeDepth browser authorization flow. Cursor uses an MCP entry with a `url`, while Codex uses an `mcp_servers` configuration and `codex mcp login edgedepth`.

For local stdio, install nothing permanently: run `npx -y @edgedepth/research-mcp` with `EDGEDEPTH_API_KEY` set to a key created on the EdgeDepth Developer page. Node.js 20 or newer is required. The `research:read` scope covers recorded-data tools; `research:interpret` is additionally needed for `interpret_prose`.

## Tools and capabilities

- Inspect feature IDs, types, ranges, operators, windows, sequence rules, limits, and error codes with `list_features`.
- Check the research universe and coverage with `list_instruments`.
- Run scans, retrieve pages, and compare matched predicates with `run_scan`, `next_page`, and `run_cohort`.
- Analyze recorded events with `ground_screenshots`, `investigate_move`, and `snapshot_at`.
- Measure prevalence, intersections, stratified populations, and outcome-first patterns with `base_rate`, `commonality`, `run_stratified`, and `outcome_first`.
- Retrieve a published report using its eight-character canonical hash with `get_report`.

## Limitations and notes

The EdgeDepth Research MCP server is research-only. Its tools do not trade, change alerts, or publish reports. Fresh scans, cohort calculations, and stratified computations may consume research allowance units, and the response identifies that side effect. `prepare_study`, `interpret_prose`, and screenshot grounding are described as free where applicable.

Outcome fields cannot be used as scan filters, which prevents lookahead selection in the supported workflow. Missing data is represented as absent rather than silently converted to zero. Page rows are examples; population counts and outcome denominators should be used for conclusions. The optional private Radar pilot can save account-owned hypotheses only with the separate `research:hypotheses` permission.

_Full upstream README: https://allmcps.com/mcp/edgedepth-research-2/readme_