Skip to content
MCP server

trading-research-agents

Read-only checks for trading claims: cost floor, dataset audit and validation records

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, self-hosted
Runs
Locally
Version
0.10.0
Listing updated
2 days ago

About trading-research-agents

trading-research-agents is an open-source MCP server for finance and payments. It runs locally on your machine and is started by your AI client.

This server exposes the research gates of TRA, a crypto trading desk you run yourself, to any agent host over stdio. The point is that it only reads. It cannot trade, promote a tier or write a ledger row, which the docs describe as a guarantee: capital is governed by a tested, fail-closed system, not by a tool a model can call.

Its ten tools answer questions such as whether a claimed edge clears the round-trip cost, whether a dataset is fit to test on, and what a validation run decided. Examples include check_cost_floor, audit_dataset for .csv or .zip archives, validate_trades against a registration, and show_ledger for the kill ledger. Paths resolve on the machine running the server, basis-point values are decimal strings and times are UTC. A refused gate comes back as a readable tool error.

The README says TRA is experimental, has not been run with real capital by its authors, and ships no profitable strategy. Install with pip install 'bitra[mcp]' and run tra mcp serve, or launch it on demand with uvx.

Install

Run in a terminal

claude mcp add trading-research-agents -- uvx trading-research-agents

Add to claude_desktop_config.json

{
  "mcpServers": {
    "trading-research-agents": {
      "command": "uvx",
      "args": [
        "trading-research-agents"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "trading-research-agents": {
      "command": "uvx",
      "args": [
        "trading-research-agents"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "trading-research-agents": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "trading-research-agents"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "trading-research-agents": {
      "command": "uvx",
      "args": [
        "trading-research-agents"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "trading-research-agents": {
      "command": "uvx",
      "args": [
        "trading-research-agents"
      ]
    }
  }
}

Tools this server provides (10)

Taken from the project documentation; the running server may differ.

check_cost_floor
Checks whether a claimed edge clears the round-trip cost.
observe_events
Reports events and response distribution from a dataset.
check_registration
Reads a preregistration and returns its seal, horizon and kill criteria.
audit_dataset
Audits a price archive and returns findings.
validate_trades
Validates trades against a registration.
show_ledger
Shows the kill ledger.
show_ladder
Shows the ladder tier and history.
read_run_report
Reads a run summary.
search_ledger
Searches ledger rows by query.
list_runs
Lists past runs, newest first.

What it does

Cost floor check
Tests whether a claimed edge in basis points clears the round-trip cost.
Dataset audit
Audits a csv, csv.gz or zip archive and returns every finding.
Trade validation
Validates a trades file against a registration with a recorded decision.
Kill ledger access
Shows ledger counts per verdict and searches rows that stand.
Run reports
Reads past runs and summarizes closed round trips.
Read-only by design
No tool sends orders, writes ledger rows or moves the ladder.

Good for

  • Check whether a strategy's claimed edge survives trading costs
  • Audit a price-data archive before backtesting
  • Review what a validation run decided and why
  • Search past rejected ideas in the kill ledger
  • Let a desktop agent inspect runs in a project folder

Pricing

Open source

Prices change. Check docs.bijectivelabs.dev before you buy.

trading-research-agents FAQ

Can it place trades?

No. Every tool is read-only and none sends an order.

How is it launched?

With tra mcp serve over stdio, or through uvx with the bitra package.

How many tools does it have?

Ten, all annotated read-only.

What data does it need?

Local files such as envelope.toml, a Parquet dataset or a trades.jsonl on the machine running the server.

Is it production ready?

The README calls TRA experimental and says it has not been run with real capital by its authors.

Is trading-research-agents free?

Yes. trading-research-agents is open source, so you can use and self-host it for free.

How do I add trading-research-agents to Claude Code?

Run this command in your terminal: claude mcp add trading-research-agents -- uvx trading-research-agents

What are the best alternatives to trading-research-agents?

The closest alternatives to trading-research-agents are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

Alternatives to trading-research-agents

  1. REA

    MCP server

    CLI and MCP server that lets coding agents reverse engineer apps and binaries

    Runs locally
    • Open source
    • 464K GitHub stars
    • Developer tools
  2. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  3. GitHub

    MCP server

    Official MCP server connecting AI assistants to GitHub repos, issues and workflows

    Not checked yet
    • Remote
    • Open source
    • 34.5K GitHub stars
    • Developer tools
  4. MCP toolkit giving coding agents symbol-level code retrieval and editing

    Runs locally
    • Open source
    • 31.5K GitHub stars
    • Developer tools
  5. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  6. Ouroboros

    MCP server

    Spec-first layer that pins requirements and verifies results of AI coding agent work

    Runs locally
    • Open source
    • 6.4K GitHub stars
    • Developer tools