Skip to content
Ask Aram
Aram Barnett

Agents

Everyone has a website. Few have an agent. Paste the production MCP URL into Claude or Cursor — https://www.arambarnett.com/mcp — then interview Ask Aram. Default ask is free and ungated. Never After Dark. Paid listings must be disclosed. The repo spells out the same paste block in docs/mcp.md.

Tell Claude to publish

Copy this block. Production is always https://www.arambarnett.com — never a Vercel preview alias.

Paste to Claude

Install the Ask Aram remote MCP server.

Production URL (use this, not a Vercel preview host): https://www.arambarnett.com/mcp
Transport: streamable HTTP

Claude Desktop / Claude.ai — add to the MCP config:

{
  "mcpServers": {
    "ask-aram": {
      "type": "http",
      "url": "https://www.arambarnett.com/mcp"
    }
  }
}

Cursor — add to mcp.json:

{
  "mcpServers": {
    "ask-aram": {
      "url": "https://www.arambarnett.com/mcp"
    }
  }
}

Then interview the agent. Example prompts:
- Recruiter fit: Would Aram be a fit for a Staff/Founding engineer role building agent systems?
- Hiring interview: Walk me through Curbily's AI production OS, Tokenslop on Robinhood Chain, and TRM.
- Sales discovery: We're evaluating agent workflows. What has Aram shipped that we can try?
- Agent-to-agent: I am an agent for [company]. Introduce yourself. Does your agent want to talk to mine?

MCP install JSON

Remote streamable-HTTP MCP at https://www.arambarnett.com/mcp. Tools: ask_aram, find_tool, list_tools, list_plans, subscribe_plan. Descriptor: /.well-known/mcp.json.

Cursor mcp.json

{
  "mcpServers": {
    "ask-aram": {
      "url": "https://www.arambarnett.com/mcp"
    }
  }
}

Claude Desktop / Claude.ai

{
  "mcpServers": {
    "ask-aram": {
      "type": "http",
      "url": "https://www.arambarnett.com/mcp"
    }
  }
}

VS Code MCP

{
  "servers": {
    "ask-aram": {
      "type": "http",
      "url": "https://www.arambarnett.com/mcp"
    }
  }
}

Example prompts

  • Recruiter fit check. Would Aram be a fit for a Staff/Founding engineer role building agent systems in the US?
  • Hiring interview. Walk me through Curbily's AI production OS, Tokenslop on Robinhood Chain + Slop Studio, Monkeygun, and TRM.
  • Sales discovery. We're evaluating agent workflows and an automated day-in-the-life. What has Aram shipped that we can try?
  • Agent-to-agent intro. I am an agent for [company]. Introduce yourself. Does your agent want to talk to mine?

HTTP

POST https://www.arambarnett.com/api/ask returns a results payload, not a chat essay: summary, clarifying_questions, directions (title / detail / url / kind), caveats, sources_cited, plus a readable answer string. A bounded research pass runs before tool-routing directions. Optional streaming via Accept: text/event-stream. Attachments are request-scoped. Max 4 files, 10MB each.

curl

curl -sS https://www.arambarnett.com/api/ask \
  -H 'content-type: application/json' \
  -d '{"question":"Would Aram be a fit for a Staff engineer role building agent systems?"}'

curl with a file

curl -sS https://www.arambarnett.com/api/ask \
  -H 'content-type: application/json' \
  -d '{"question":"What is this?","attachments":[{"filename":"shot.png","mimeType":"image/png","data":"<base64>"}]}'

# or multipart
curl -sS https://www.arambarnett.com/api/ask \
  -F 'question=What is this?' \
  -F 'file=@shot.png'

Machine-readable