Open Ethics Standard MCP

A shared ethical standard for any business. The Open Ethics Standard (OES) says how a business should treat the people it serves, employs, buys from, and affects, and names the specific ways that goes wrong. It covers every industry: hotels, gyms, lenders, employers, manufacturers, and software.

The eight principles

Principle	In short
Honesty	Tell people the truth about prices, products, services, and claims.
Autonomy	People choose freely and knowingly, and can say no, change their mind, or leave.
Privacy	Collect only what is needed, use it only as agreed, and protect it.
Fairness	Treat people equally, do not discriminate, and keep services accessible.
Care	Protect vulnerable people; do not profit from distress, grief, or compulsion.
Safety	Do not sell known harms, keep service promises, and allow repair.
Accountability	Explain decisions, offer recourse, own mistakes, and protect people who speak up.
Responsibility	Workers, suppliers, communities, competitors, and the environment count too.

Under them sit 804 named, numbered practices that break these principles, such as Drip Pricing (§ 1.1.1) or Roach Motel Friction (§ 2.1.3). Read the standard.

Ways to use it

• Cite it. Name a harm precisely, with a number anyone can look up: "the cancellation flow breaks Autonomy, OES § 2.1.1."
• Assess a business. Work through each principle and the practices under it. Each practice has a plain description and examples from many industries.
• Apply it to software. The OES MCP lets your AI coding assistant review a project against the standard. It runs on your machine. Findings are leads for you to review, not verdicts or certifications, and a code review can only see what a product does, not how a company treats its workers or suppliers.

Apply it to your software

1. Add the MCP. It needs Node.js 20+; no account or API key:

claude mcp add --scope user open-ethics-standard -- npx -y open-ethics-standard-mcp@latest

For Codex: codex mcp add open-ethics-standard -- npx -y open-ethics-standard-mcp@latest. For Cursor, VS Code, or other clients, add a stdio server with command npx and args -y open-ethics-standard-mcp@latest (per-client commands). On native Windows, use cmd /c npx …. To uninstall: claude mcp remove --scope user open-ethics-standard.

2. Ask for a review. Open your project and say:

Review my project against the Open Ethics Standard.

You get back one link to an offline report, with every finding a click away:

How it works

1. Your AI reads your code. The MCP gives it the principles, the catalog, and a review workflow.
2. It matches behavior to OES practices, citing files, lines, evidence, and the principle each one breaks.
3. You decide. Open the report, mark each finding Applies, Unsure, or Dismiss, and export to Markdown or PDF. After fixes, ask for a recheck; the same report tracks what's fixed.

Your data

• The MCP runs on your machine. It never sends your code or findings anywhere, and your AI never has to download or run scripts to build the report.
• Reports are saved in ~/.local/share/open-ethics-standard/reports (override with OES_REPORT_DIR).
• @latest picks up new catalog releases automatically. Pin a version ([email protected]) if you'd rather update by hand.

Contribute

Spotted a practice that's missing or wrong? Edit OPEN-ETHICS-STANDARD.md and open a pull request, or open an issue. See CONTRIBUTING.md.

License

Code is MIT. The standard text and catalog are CC BY-ND 4.0.