Built my personal MCP server

Main outcome of the day: I built my own MCP server and published it for direct use.

IBX parsed snapshot for 2026-04-27

Pulled from IBX CLI:

• done tasks: 4
• open tasks due that day: 0
• done priority split: 3 priority-1, 1 priority-2, 0 priority-3
• estimated effort completed: 3.75 hours

MCP server

Public endpoint:

• /mcp
• https://egeuysal.com/mcp

If anyone wants to connect it in Codex or Claude, these docs are the references:

• OpenAI Codex MCP docs
• Claude Desktop local MCP guide

Tools available (to my current knowledge)

{
  "name": "egeuysal-www",
  "transport": "MCP streamable HTTP",
  "endpoint": "/mcp",
  "protocol": {
    "method": "POST",
    "headers": {
      "accept": "application/json",
      "content-type": "application/json"
    },
    "exampleBody": {
      "id": 1,
      "jsonrpc": "2.0",
      "method": "tools/list"
    }
  },
  "tools": {
    "get_diary_json": {
      "description": "Return this site's /diary.json feed.",
      "arguments": {}
    },
    "get_blog_json": {
      "description": "Return this site's /blog.json feed.",
      "arguments": {}
    },
    "get_agents_json": {
      "description": "Return this site's /agents.json feed.",
      "arguments": {}
    },
    "get_photo_json": {
      "description": "Return this site's /photo.json feed.",
      "arguments": {}
    },
    "search_brain_resources": {
      "description": "Search brain.egeuysal.com/api/routes.json across paginated route pages. Returns docs with markdownUrl.",
      "arguments": {
        "query": "optional text search over title/path/sourcePath",
        "routePrefix": "optional prefix such as resources/articles or resources/posts",
        "limit": "optional 1-200, default 50",
        "maxPages": "optional 1-100, default 50"
      }
    },
    "fetch_brain_resource": {
      "description": "Fetch markdown from brain.egeuysal.com. Accepts markdownUrl, page URL, routePath, or /resources path.",
      "arguments": {
        "urlOrPath": "for example https://brain.egeuysal.com/resources/articles/2026-03-21-133927.md"
      }
    },
    "get_bri_notes": {
      "description": "Fetch Bri notes from bri.fyi/api/notes.",
      "arguments": {
        "query": "optional text search over title/slug/content",
        "limit": "optional 1-200, default 50"
      }
    },
    "get_ibx_todos": {
      "description": "Fetch IBX todos from ibx.egeuysal.com/api/todos.",
      "arguments": {
        "today": "optional YYYY-MM-DD, defaults to local America/Chicago date",
        "status": "optional all/open/done, default all",
        "query": "optional text search over title/notes",
        "limit": "optional 1-200, default 50"
      }
    }
  }
}

Takeaway

This was a focused builder day. The main win is shipping the MCP surface so both Codex and Claude can query my real data sources from one place.