Boikot MCP Server for Company Ethics Data

Overview

Boikot MCP Server exposes a lightweight Model Context Protocol (MCP) endpoint that provides programmatic access to a community-curated dataset of company ethics records. The project collects and documents ethical and unethical corporate behavior so developers, researchers, and consumer tools can surface transparent context about companies when making decisions or building AI assistants.

The MCP server is publicly available (no authentication) and is designed to be easy to integrate into downstream services: it offers a single lookup tool that returns structured company ethics metadata, source links, tags, logos and timestamps. The underlying dataset is distributed as a single JSON file (boikot.json) under the GPL v3 license, enabling reuse and redistribution consistent with that license.

Features

• Public MCP endpoint at https://mcp.boikot.xyz/mcp (no auth)
• Single tool: lookup_company_information(company_name)
• Returns structured company ethics records: names, comment, tags, sources, logo and site URLs, updatedAt
• Community-curated dataset available as boikot.json (raw GitHub link)
• Open-source (GPL v3) and intended for integration with research, consumer tools, and AI agents

Installation / Configuration

Clone the repo and run the included script to start the site and a local backend.

# clone the repo
git clone https://github.com/boikot-xyz/boikot.git
cd boikot

# run the bundled runner (starts frontend + local backend)
./run.sh

Notes:

• The website frontend is a React app in the site directory.
• Additional data collection and backend utilities live in scripts and backend.
• The run.sh script described above is the simplest way to run the project locally; no further configuration is required for basic development.

How to query the MCP server

The MCP server exposes a tool named lookup_company_information that accepts a single parameter company_name. The server requires no authentication. The following example demonstrates a generic POST request format that many MCP-compatible tools accept; adapt to your client or framework as needed.

Example curl (generic JSON POST):

curl -X POST "https://mcp.boikot.xyz/mcp" \
  -H "Content-Type: application/json" \
  -d '{"tool": "lookup_company_information", "company_name": "Example Corp"}'

Example simplified response (fields described below):

{
  "names": ["Example Corp", "Example Corporation"],
  "comment": "Reported to have labor violations in region X [1][2].",
  "sources": {
    "1": "https://example.org/report1",
    "2": "https://example.org/report2"
  },
  "tags": ["labor", "environment", "controversy"],
  "logoUrl": "https://example.com/logo.png",
  "siteUrl": "https://example.com",
  "updatedAt": "2024-03-01T12:34:56Z"
}

Available Resources

• boikot.json (full dataset): https://raw.githubusercontent.com/boikot-xyz/boikot/main/boikot.json
• MCP endpoint: https://mcp.boikot.xyz/mcp
• GitHub repository: https://github.com/boikot-xyz/boikot
• License: GPL v3 (see repository for details)

Field reference

Use Cases

• Chatbots & virtual assistants: Enrich assistant responses with concise, sourced company ethics context. Example: when a user asks “Is it ethical to buy from X?”, call the MCP tool and include the comment and sources in the reply.
• Browser extensions / shopping apps: Display ethics badges or warnings on product pages by querying the MCP for the merchant or brand name and surfacing tags and comment.
• Research & dashboards: Aggregate company records locally from boikot.json to build visualizations of ethical trends across sectors, or to cross-reference with procurement lists.
• Automated monitoring: Integrate into pipelines that watch news feeds or social media; when a company name appears, call the lookup tool to tag the mention with known ethics data and link to primary sources.

Tips for Developers

• Prefer matching on canonical names from the names array; the first entry is the primary label.
• Use sources to cite original reporting when presenting information to end users.
• Respect GPL v3 when redistributing the dataset or derived services.
• If running locally for development, use the included run.sh to avoid manual environment setup.

For code-centric integrations or agent workflows, treat the MCP endpoint as a deterministic tool that returns structured company records you can programmatically inspect, index, or render.