mcpField
V
VolatiAI MCP

VolatiAI MCP

Remote
github.com/christianscode1994Data & Analytics0Updated today
Repository

About VolatiAI MCP

VolatiAI analyzes correlations across sector time series from public-domain data, attaching uncertainty, provenance, and stability information to each claim. It is for analysts interpreting cross-sector signals and explicitly reports empty or unsupported results.

Key features

  1. 01Analyze pairwise correlations of detrended residuals
  2. 02Provide multiple-testing and permutation corrections
  3. 03Report correlation confidence intervals
  4. 04Classify provenance and stability
  5. 05Use named verdicts to report claim status

Use cases

  1. 01Assess sector relationships in public data
  2. 02Check correlations for multiple-testing correction
  3. 03Distinguish same-quantity checks from cross-sector insights
  4. 04Review daily signal stability and no-evidence results

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://volatiai-mcp.fourieranalys.workers.dev/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

VolatiAI

Correlation analysis of public-domain data. Every claim carries its own uncertainty.

Live: https://christianscode1994.github.io/VolatiAI/

What it does

VolatiAI collects ~500 sector time-series from public-domain sources (US government, EU open data, CC0), computes pairwise correlations on detrended residuals, and reports every claim with:

  • p-value and Benjamini-Hochberg adjusted p
  • Permutation p (assumption-free)
  • Fisher 95% CI on the correlation coefficient
  • Provenance classification (same-quantity vs. cross-sector insight)
  • Stability score across days
  • Named verdict: confirm:provenance, surface, suppress:, dispute:

When the system cannot verify something, it returns that fact explicitly (empty transitivity, empty drifts, no_evidence verdicts). Empty results are information, not absence.

Live endpoints

Repository layout

workers/sector-slow/src/ — Worker source (Cloudflare Workers) index.js — Collectors, endpoints, statistical honesty layer arbitration.js — Pure statistical functions (pearson, detrend, arbitrate)

scripts/ — Data pipeline briefing.py — Generates docs/briefing.json from /signal data find_candidates.py — Provenance candidate detection

tests/ — Unit tests (Python + Node) test_briefing.py — 17 tests test_find_candidates.py — 10 tests arbitration.mjs — 20 tests correlation_pvalue.mjs — 6 tests documenting correlationPValue behavior

docs/ — Static site (GitHub Pages) index.html — Dashboard methodology.html — Field-by-field explanation api.html — Interactive API reference

config/ — Registries active_sectors.json — Sector registry semantic_registry.json — Same-quantity groups provenance.json — Per-sector attribution

Running tests

python3 tests/test_briefing.py
python3 tests/test_find_candidates.py
node tests/arbitration.mjs
node tests/correlation_pvalue.mjs

Design principles

  1. Honest empty results. When there is nothing to report, say so explicitly.
  2. Public domain only. Every data source is US government, EU open data, CC0, or CC-BY.
  3. Multiple-testing correction is mandatory. BH-FDR at q=0.05 is the floor.
  4. Provenance is a control, not an insight. Same-quantity pairs validate the pipeline.
  5. Log-rate, not raw percentage. Window length should not distort comparisons.

Status

  • Worker: deployed with 5 cron shards (03:17-03:57 UTC daily)
  • Tests: 105 across 8 files, all passing locally
  • Snapshot signing: Ed25519 module ready (workers/sector-slow/src/signing.js) with 15/15 tests. Wiring Nov 8 (needs 3+ days of cron data). Verification: scripts/verify_signature.py

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.