MODEL CONTEXT PROTOCOL

Connect Sharebench to your agent.

Add the public bench to Claude Code, Cursor, ChatGPT, or Codex over MCP. One connection — then your agent can search and pull every skill, agent, prompt, and playbook on the bench, on demand. No per-artifact install, no signup.

Browse the bench

Endpoint https://mcp-public.sharebench.ai · transport streamable-http · read-only · no auth

Set it up

Pick your tool and paste. (Or hit Connect via MCP above for a copy-button walkthrough.)

Claude Code

# Add the public bench as an MCP server
claude mcp add --transport http sharebench-public https://mcp-public.sharebench.ai

Cursor

# Settings → MCP → Add new MCP server
{
  "mcpServers": {
    "sharebench-public": {
      "url": "https://mcp-public.sharebench.ai"
    }
  }
}

ChatGPT

# Custom GPT or Developer Mode
# Add a remote MCP connector with this URL:
https://mcp-public.sharebench.ai

Codex CLI

# Append to ~/.codex/config.toml
[mcp_servers.sharebench_public]
url = "https://mcp-public.sharebench.ai"

What your agent gets

search — find skills, agents, prompts, playbooks & workflows by name or natural-language concept (hybrid lexical + semantic ranking).
get_by_id — pull the full SKILL.md content of any artifact straight into the agent's context.
Read-only and free — the public bench runs on a read-only role, so connecting is safe and needs no account.

In the official MCP Registry

Sharebench is published as ai.sharebench/registry — so MCP-aware clients can discover it automatically. Verify the listing →