mcpField
J
JW3 MCP Server

JW3 MCP Server

Remote
jw3.aiFinance0Updated today
RepositoryVisit website

About JW3 MCP Server

JW3 MCP Server provides remote market-reading tools and trading execution with confirmation gates. It is for clients that need access to JW3.ai market data and controlled execution; tool calls require OAuth sign-in or a scoped PAT.

Key features

  1. 01Remote Streamable HTTP access
  2. 02Market-reading tools
  3. 03Confirmation-gated trade execution
  4. 04OAuth and scoped PAT authentication

Use cases

  1. 01Read market data through an MCP client
  2. 02Execute trades with confirmation
  3. 03Authenticate tool calls with OAuth or a scoped PAT

MCP support

It is a remote MCP server using Streamable HTTP. It exposes market reads and confirm-gated trading execution; discovery is open, while tool calls require OAuth or a scoped PAT.

Configuration

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

Endpointhttps://nzhbfeahymmyqerzjgbt.supabase.co/functions/v1/mcp-server
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

JW3 MCP + CLI

mcptry

Listed in the official MCP Registry as io.github.agnostic4321-ops/jw3-mcp-cli (remote, Streamable HTTP). Discovery (initialize, tools/list) is open; every tool call needs OAuth sign-in or a scoped PAT.

Standalone harness repo: the MCP tool server (Supabase edge functions) and the jw3 command-line thin wrapper over its JSON-RPC surface.

Split out of jw3_ai (2026-10-26) to keep app and harness lifecycles separate. The deployed function URLs are unchanged — same Supabase project, same slugs.

Layout

  • cli/ — @jw3/cli (jw3 / jw3d bins). Pure HTTP client of the mcp-server surface; no app imports. pnpm --filter @jw3/cli build/test.
  • edge-functions/ — Supabase edge functions, deployed to the same project:
    • mcp-server (gateway JWT off — harness PATs are the gate)
    • mcp-auth, mcp-token, mcp-oauth (token issuance / OAuth)

Deploying functions

# Syncs edge-functions/{_shared,agent-chat} from the ../jw3_ai checkout
# (override with JW3_AI_DIR), runs deno check, then stages the standard
# supabase/functions layout in a temp dir and deploys.
scripts/sync-edge-shared.sh --deploy mcp-server

Direct supabase functions deploy from this root does NOT work: sources live under edge-functions/ (not supabase/functions/), and the ../_shared + ../agent-chat imports only resolve after the sync step above (previously a throwaway-symlink step — now this committed script; the copies are gitignored, never committed). Sync-only (no deploy): run the script with no args.

supabase/config.toml pins the gateway flags (verify_jwt = false where the harness PAT — not a JWT — is the credential).

Parity contract

jw3_ai/src/test/read-tool-parity.test.ts reads this repo at the sibling path ../jw3-mcp-cli/edge-functions/mcp-server/* when present and skips otherwise. Keep read_live_datum kinds and shared readers (fetchOnchainSnapshot, executeScreenerRead, no inline buildOnchainAggregate) in sync with jw3_ai's agent-chat/read-tools.ts — the test enforces it locally.

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.