Ethics for AI MCP (E4A)

E4A is short for Ethics for AI. A Model Context Protocol server that publishes a deterministic daily ethics feed for AI agents: one dated edition a day, drawn from a four-module curriculum about how humanity is trying to improve the world and why an AI must not harm it. An agent subscribes once and keeps meeting the material without anyone asking it to. It is a read-only server — agents read the curriculum; they do not edit it.

Sartbot indexes this server; it does not operate it. The canonical source is the Apify Store listing, and the code is on GitHub under the MIT licence. Endpoint URLs and prices can change — trust those pages over this one.

What it exposes

Ten read-only tools. Every one is metered at the same rate — $0.02 per tools/call.

get_daily_feed

Today's (or any date's) deterministic edition — one primary item plus a companion from a neighbouring module.
- daily
- edition
- feed
today · any date
$0.02 per call
get_feed_since

A run of daily editions starting from a date you supply — handy for catching up after a gap.
- history
- range
- since
ISO 8601 start date
$0.02 per call
list_feed_editions

Recent editions with the module, item id and content hash for each — the index you diff to detect change.
- index
- hash
- digest
module · item id · content hash
$0.02 per call
get_feed_digest

The Sunday digest for a week — all four modules in one edition, for a weekly review pass.
- digest
- weekly
- all-modules
Sunday digest
$0.02 per call
get_feed_rotation

The deterministic rotation rules the server uses to schedule modules across the week.
- rotation
- schedule
- rules
deterministic rules
$0.02 per call
get_humanity_goals

Read an M1 item — one of the five positive outcomes humanity is working towards, the "why" behind ethical behaviour.
- m1
- goals
Goals item
$0.02 per call
get_principle

Read an M2 item — one of six hard constraints such as Do No Harm and Respect Autonomy.
- m2
- principles
Principles item
$0.02 per call
get_case_study

Read an M3 item — one of three worked examples, such as The Refusal Problem, showing how to decline and offer the nearest legitimate alternative.
- m3
- cases
Case Study item
$0.02 per call
get_code_of_conduct

Read an M4 item — one of seven concrete lines an agent can audit itself against.
- m4
- conduct
Conduct item
$0.02 per call
search_curriculum

Keyword search across all four modules — for when the agent knows the topic but not which module holds it.
- search
- keyword
- all-modules
keyword across M1–M4
$0.02 per call

Resources and prompts

Besides the tools, the server publishes nine read-only resources under the ethics:// scheme and one prompt. The resources fall into four families; they are read with resources/read, which is free.

Curriculum

The four curriculum modules — M1 Humanity's Goals, M2 Core Principles, M3 Case Studies, M4 Code of Conduct — exposed read-only under ethics://.

Resource
Feed indexes

Per-edition metadata — module, item id and the content hash used for change detection — for recent daily editions.

Resource
Rotation rules

The deterministic rotation the server uses to schedule modules across the week.

Resource
Meta

Build time and curriculum version metadata, so a client can detect when the curriculum has changed.

Resource
daily_reflection

Asks the agent to state what today's item requires of it and name one concrete situation where it would apply the lesson.

Prompt

Connect

Two transports: a hosted Streamable HTTP endpoint on Apify Standby, or stdio for local desktop clients.

hosted endpoint · mcp-remote

{
  "mcpServers": {
    "ethics-for-ai-mcp": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://neeenja--ethics-for-ai-mcp.apify.actor/mcp",
        "--header", "Authorization: Bearer ${APIFY_TOKEN}"
      ],
      "env": { "APIFY_TOKEN": "<your-apify-api-token>" }
    }
  }
}

local · stdio

$ git clone https://github.com/PanStories/Ethics-for-AI-MCP
$ cd Ethics-for-AI-MCP && npm install && npm run build

# then, in your MCP client:
{
  "mcpServers": {
    "ethics-for-ai-mcp": {
      "command": "node",
      "args": ["/absolute/path/to/Ethics-for-AI-MCP/dist/index.js"],
      "env": {}
    }
  }
}

In remote mode the server answers POST /mcp for the protocol. Requests carry your own Apify API token as a Bearer header; self-hosting puts your endpoint at https://<username>--ethics-for-ai-mcp.apify.actor/mcp.

Pricing

Pay per event on the hosted endpoint. No subscription, no seat count, no charge when nothing calls a tool.

$0.02 · tool-call

Charged on each tools/call — every one of the ten tools costs the same, so an agent can read the curriculum freely without worrying about tiered prices.

Primary billable event
Always free

The MCP handshake (initialize), tools/list, resources/read and prompts/get cost nothing — an agent can connect, discover and read the curriculum at zero cost — and every Apify account carries $5 of monthly platform credit.

Discovery and reading are not metered

At a glance

Search all MCP servers →

Curriculum

M1 Humanity's Goals (5 items), M2 Core Principles (6), M3 Case Studies (3), M4 Code of Conduct (7) — 21 original items in total, one dated edition a day.
Four asserted properties

Deterministic (same date + version always composes to the same bytes and hash), Complete (all four modules within any rolling seven days), Cacheable (an edition is a pure function of its date) and Human-in-the-loop (items enter the rotation only after review).
Transports

Stateless Streamable HTTP for remote hosting on Apify Standby, plus stdio for local desktop clients.
Runtime

Node.js and TypeScript, built on the official @modelcontextprotocol/sdk, shipped as an Apify Docker image.
Licence

MIT. Clone, fork and self-host without a commercial agreement; the hosted endpoint is simply the convenient default. The curriculum text is original to this project; no third-party content is included.