AI & Machine Learning

poorpaul.dev

Provides tools to explore GPU benchmark data and recommend quantization strategies for AI models.

1 endpoint14 known toolsFirst detected August 12, 2026Last detected August 12, 2026

ENDPOINT 1

https://mcp.poorpaul.dev/mcp

No auth detected

MCP server metadata

Name: Poor Paul's MCP
Version: 0.1.0

Capabilities

experimentalloggingprompts.listChangedresources.listChangedtools.listChangedextensions.io.modelcontextprotocol/ui

Server instructions

Queryable GPU inference benchmarks from Poor Paul's Benchmark (PPB). Quantitative tools: recommend_quantization, query_ppb_results, get_gpu_headroom, list_tested_configs, compare_quants_quantitative, get_combined_scores, rank_by_priority, recommend_hardware, explain_result. Qualitative tools: get_qualitative_summary, query_qualitative_results, get_context_rot_breakdown, get_tool_accuracy_breakdown, compare_quants_qualitative. Data source: https://huggingface.co/datasets/paulplee/ppb-results

Known tools 14

list_tested_configs

List every tested GPU, model, quantization, and runner type in the PPB dataset.

Inferred read-only

query_ppb_results

Filter raw benchmark rows from PPB.

Inferred read-only

recommend_quantization

Recommend the best quantization for a given GPU VRAM budget and user count.

Inferred read-only

get_gpu_headroom

Sanity-check VRAM usage for a specific (gpu, quant, model, users) config.

Inferred read-only

compare_quants_quantitative

Compare quantitative benchmark scores across quantizations for a model.

Inferred read-only

get_combined_scores

Get both quantitative (speed/VRAM) and qualitative (accuracy/quality) scores for a single (gpu, model, quant) configuration in one call.

Inferred read-only

rank_by_priority

Rank all tested quantizations for a model by a composite score.

Inferred read-only

recommend_hardware

Recommend the best GPU for running a specific model.

Inferred read-only

explain_result

Explain why a benchmark result is what it is.

Inferred read-only

get_qualitative_summary

Get all available qualitative benchmark scores for a model, optionally filtered to a specific quantization and/or GPU.

Inferred read-only

query_qualitative_results

Query qualitative benchmark results with optional filters.

Inferred read-only

get_context_rot_breakdown

Get detailed context-rot (long-context recall) scores broken down by context length, insertion depth, and needle type.

Inferred read-only

get_tool_accuracy_breakdown

Get detailed tool-call accuracy metrics for a model+quant.

Inferred read-only

compare_quants_qualitative

Compare qualitative benchmark scores across quantizations for a model.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.poor-paul-s-mcp]
url = "https://mcp.poorpaul.dev/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "poor-paul-s-mcp": {
      "type": "http",
      "url": "https://mcp.poorpaul.dev/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: poor-paul-s-mcp
Remote MCP URL: https://mcp.poorpaul.dev/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "poor-paul-s-mcp": {
      "url": "https://mcp.poorpaul.dev/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "poor-paul-s-mcp": {
      "type": "http",
      "url": "https://mcp.poorpaul.dev/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "poor-paul-s-mcp",
  "transport": "streamable-http",
  "url": "https://mcp.poorpaul.dev/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

TRUST AND VERIFICATION EVIDENCE

Trust Data Available

BuiltWith Trust API v2 evidence for poorpaul.dev was fetched 2026-08-23T12:34:41.974Z.

Trust status Neutral

poorpaul.dev is assessed as Neutral: No suspicious signals found, but no strong positive signal either

Indexed

Evidence is source-attributed and does not guarantee that a third-party server is safe. Risk labels are conservative metadata heuristics.