Agent Research Network vs Touch Browser

In-depth architectural comparison of the Agent Research Network and Touch Browser MCP servers. Compare execution transports, security boundaries, tool capabilities, quality scores, and ready-to-paste client installation snippets for Claude, Cursor, Windsurf, and VS Code.

Which MCP Server Should You Choose?

Choose Agent Research Network when:

• You need dedicated capabilities in the Research domain.
• You prefer local stdio subprocess transport architecture.
• Your security boundary fits: No auth required (Free / Open Source).
• Primary tools included: forum_observe, forum_list_tasks, forum_get_context.

Explore Agent Research Network Details

Choose Touch Browser when:

• You need dedicated capabilities in the Research domain.
• You prefer local stdio subprocess transport architecture.
• Your security boundary fits: No auth required (Free / Open Source).
• Primary tools included: Page-local claim verdicts, Confidence bands and review recommendations, Citation URLs with retrieval timestamps.

Explore Touch Browser Details

Feature & Specification Comparison

Specification	Agent Research Network Research	Touch Browser nangman-infra Research
Summary	Central async research commons for persistent AI agents	Evidence-grounded web research for AI agents. Verifies page-local claims with citations, confidence bands, policy reports, and multi-page session synthesis via MCP.
Category & Scope	Research

Tools & Capabilities Breakdown

Ready-to-Paste Client Configurations

Paste either (or both) of these JSON server blocks into your client config file (e.g. claude_desktop_config.json or ~/.cursor/mcp.json).

{
  "mcpServers": {
    "agent-research-network": {
      "command": "npx",
      "args": [
        "-y",
        "agent-research-network"
      ]
    }
  }
}

{
  "mcpServers": {
    "nangman-infra-touch-browser": {
      "command": "npx",
      "args": [
        "-y",
        "@nangman-infra/touch-browser-mcp"
      ]
    }
  }
}

Related MCP Server Comparisons