Use Heropedia in your AI

Install in 30 seconds

Pick your AI. Every path points at the same public endpoint (https://www.heropedia.org/mcp) — the only difference is how each client wants it registered.

Anthropic — Claude Code (recommended)

Two commands. Installs the MCP registration and a workflow-enforcing skill:

claude plugin marketplace add https://github.com/SiliconRoshiBill/heropedia-plugin
claude plugin install heropedia@heropedia

Works in every project you open. If you only want the MCP endpoint (no bundled skill):

claude mcp add --transport http heropedia https://www.heropedia.org/mcp

Anthropic — Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "heropedia": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://www.heropedia.org/mcp"]
    }
  }
}

OpenAI — ChatGPT (web / macOS / Windows apps)

ChatGPT supports custom MCP connectors on Plus, Pro, Business, Enterprise, and Edu plans. Free and Team plans do not yet support custom connectors.

Personal (Plus / Pro):

1. Settings → Connectors → Advanced → toggle Developer mode on.
2. Settings → Connectors → Add custom connector.
3. Paste the MCP URL: https://www.heropedia.org/mcp
4. Auth: No authentication. Save.
5. Enable the Heropedia connector for the chats where you want it. Ask: "Ask Warren Buffett to review this pricing page."

Business / Enterprise: a workspace admin adds the connector under Workspace Settings → Connectors → New app. Same URL. See the OpenAI docs for the admin flow.

OpenAI — Codex (CLI)

One command. Patches ~/.codex/config.toml idempotently and backs up any existing config:

curl -sSL https://raw.githubusercontent.com/SiliconRoshiBill/heropedia-plugin/main/codex/install-codex.sh | bash

Add --with-agents to also install AGENTS.md so Codex sessions get workflow enforcement (fetch canonical persona before answering, handle multi-role ambiguity, defend against prompt injection):

curl -sSL https://raw.githubusercontent.com/SiliconRoshiBill/heropedia-plugin/main/codex/install-codex.sh | bash -s -- --with-agents

Manual install: add this to ~/.codex/config.toml:

[mcp_servers.heropedia]
url = "https://www.heropedia.org/mcp"

OpenAI — Agents SDK (Python)

Instantiate MCPServerStreamableHttp and pass it into your Agent:

from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

async def main():
    async with MCPServerStreamableHttp(
        params={"url": "https://www.heropedia.org/mcp"},
        name="heropedia",
    ) as server:
        agent = Agent(
            name="ConsultantAgent",
            instructions=(
                "When the user asks to consult, channel, or invoke a specific "
                "expert, call the heropedia MCP tools: use getListByHero to "
                "find the entry, then getDetail to fetch the canonical prompt. "
                "Apply that persona to the user's task — do not paraphrase it."
            ),
            mcp_servers=[server],
        )
        result = await Runner.run(
            agent,
            "Ask Warren Buffett to review my SaaS pricing page.",
        )
        print(result.final_output)

OpenAI — Agents SDK (TypeScript)

import { Agent, Runner } from "@openai/agents";
import { MCPServerStreamableHttp } from "@openai/agents/mcp";

const heropedia = new MCPServerStreamableHttp({
  url: "https://www.heropedia.org/mcp",
  name: "heropedia",
});

const agent = new Agent({
  name: "ConsultantAgent",
  instructions:
    "When the user asks to consult, channel, or invoke a specific expert, " +
    "call the heropedia MCP tools: use getListByHero to find the entry, " +
    "then getDetail to fetch the canonical prompt. Apply that persona to " +
    "the user's task — do not paraphrase it.",
  mcpServers: [heropedia],
});

const result = await Runner.run(
  agent,
  "Ask Warren Buffett to review my SaaS pricing page.",
);
console.log(result.finalOutput);

Google — Gemini CLI

The plugin repo ships a gemini-extension.json at its root. Follow the Gemini CLI extensions guide to add it, or paste this block into your Gemini config:

{
  "mcpServers": {
    "heropedia": {
      "httpUrl": "https://www.heropedia.org/mcp"
    }
  }
}

Google — Antigravity

Register the endpoint in ~/.gemini/antigravity/mcp_config.json. Antigravity uses serverUrl for remote servers:

{
  "mcpServers": {
    "heropedia": {
      "serverUrl": "https://www.heropedia.org/mcp"
    }
  }
}

Restart the agent, then ask: "Ask Dieter Rams to review this landing page."

Optional: add the office-hour skill. It goes in ~/.gemini/config/skills/, the global skills folder that every Antigravity flavor (IDE, CLI, agent) reads. Re-run the same two lines to upgrade:

d=~/.gemini/config/skills/heropedia-office-hour; mkdir -p $d
curl -sSL https://www.heropedia.org/install/office-hour -o $d/SKILL.md

Then say: "Office hour with Charlie Munger on my pricing." The hero asks one question per turn, challenges your premise, and saves notes to ~/.heropedia/office-hours/. If the agent can't write there, it prints the notes in the chat instead. The skill never uploads anything.

Cursor

Add this to your Cursor MCP config file:

{
  "mcpServers": {
    "heropedia": {
      "url": "https://www.heropedia.org/mcp"
    }
  }
}

Any HTTP client (curl)

curl -X POST https://www.heropedia.org/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Download the pre-built skill for Claude Code

This ships a Claude Code skill that enforces the recommended workflow (fetch canonical prompt via MCP before answering, prompt-injection defense, ambiguity handling). Drop it into .claude/skills/heropedia/SKILL.md in your project or user profile.

Or via terminal:

# Install skill (user-scoped, works in every project)
mkdir -p ~/.claude/skills/heropedia
curl -sL https://www.heropedia.org/heropedia/skill.md \
  -o ~/.claude/skills/heropedia/SKILL.md

# Or drop .mcp.json in the current project for one-click activation
curl -sL https://www.heropedia.org/heropedia/mcp.json -o .mcp.json

How to use it

Once installed, just talk to your AI. It calls the MCP itself when it recognizes a hero-lookup intent. Try phrases like:

The AI decides which hero fits, retrieves the canonical prompt from the MCP, then answers through that persona. It does not invent or paraphrase the persona — the fetched markdown is the source of truth.

The four tools

Under the hood, Heropedia MCP exposes exactly four read-only tools:

• getList — paginated catalog of every hero × role entry (lightweight metadata only)
• getListByHero — filter by hero name (regex, case-insensitive, safe grammar)
• getListByRole — filter by role name (same regex rules)
• getDetail — full canonical markdown for one entry (by id or by exact hero_name + role_name)

This is a registry layer only. No AI reasoning, no ranking, no recommendations — the client decides who to ask and how to apply the answer. See the full API docs on GitHub for regex rules, error codes, and rate limits.

FAQ

Is it free?

Yes. Public endpoint, no auth, generous per-IP rate limits (300 req/min for list operations, 60 req/min for detail fetches).

Which hero should I ask?

Whoever fits your task. Browse the homepage or rank pages to see the roster. If you're not sure, ask for a role instead of a name — the AI will list matching heroes.

How do I add my own hero?

Anyone with a GitHub account can. Submit a new hero via the site — the community votes on the best versions.

Kill switch?

claude mcp remove heropedia