LLM Checker MCP Server

Community

Choose and manage local LLMs for your hardware.

How to install ↓ Compare GitHub source repository ↗

Category: Dev Tools #258 of 438
Stars: ★ 3.0k Very popular
Transport: stdio (local process)
Runtime: Node.js 18+
Credentials: No credential needed
License: Other / unspecified
Last commit: 9 days ago
Tools: 28

50FMRS · D

The source describes an MCP server that exposes LLM Checker's hardware analysis, deterministic model recommendations, Ollama management, benchmarking, policy auditing, and model-file structural validation. It is suited to local model workflows; the source does not establish that it is an official server from Ollama, MCP, or another upstream product owner.

Strongest · Documentation 14/20 Weakest · Reliability 7/20

Reliability

7/20

Security and permissions

8/20

Maintenance

8/20

Documentation

14/20

Setup experience

13/20

Why each score

Reliability 7/20: The README describes MCP tools, CLI commands, exit codes, and some fail-closed behavior, but provides no server manifest, source files, tests, CI, or reviewable evidence for initialization and tool-list handshakes. Declared behavior cannot be statically verified, so the score is conservative and reduced for missing execution evidence.
Security and permissions 8/20: The README discloses local hardware inspection, Ollama management, network synchronization, and model execution. It also lists side-effecting or locally sensitive capabilities such as model removal, allowlisted CLI execution, and project scanning, but gives no verifiable permission boundaries, confirmation requirements, data-flow details, or credential-handling policy. No explicit red-line risk from the prompt is established, but the unverified privilege and confirmation controls warrant substantial deductions.
Maintenance 8/20: The repository is not archived, the README mentions multiple version highlights and release channels, and there are five open issues. However, there is no commit history, release chronology, dependency-update evidence, security-response process, or verifiable maintenance cadence. The metadata says NOASSERTION while the README claims NPDL-1.0, creating license evidence inconsistency. This does not support a high maintenance score.
Documentation 14/20: The README covers installation, CLI/MCP configuration, tool inventory, examples, runtime limitations, model-verification limits, policy workflows, and some troubleshooting. It is relatively comprehensive. Points are deducted because the server manifest is absent and source-backed details for tool parameters, authentication, permissions, network access, cost, and actual error behavior are unavailable; license information is also inconsistent.
Setup experience 13/20: The README provides npm installation, an npx option, mcp-setup commands for several mainstream clients, and configuration paths. The connection path is therefore plausible and fairly clear. Without the manifest, actual configuration contents, initialization-handshake evidence, or CI/tests, connection reliability and cross-platform behavior remain unverified, so the score is below the top range.

Static review · not runListed 2026-08-14

Read the FMRS scoring method →

Fit and risk

What it can accessRuns commands or codeUses the networkConnects to a database

Best for

Developers who run LLMs locally
Users needing hardware-aware model selection and Ollama management
Teams requiring deterministic scoring, calibrated routing, or policy audits
Users who want structural checks before running downloaded models

Not for

Users without Node.js or local Ollama
Cloud-hosted inference workflows rather than local inference
Complete guarantees about model behavior, provenance, or poisoned weights
Arbitrary system command execution beyond the MCP allowlist

Required permissions

Read local CPU, GPU, memory, and acceleration information
Access local Ollama model directories and the running Ollama environment
Depending on the tool, download, run, or remove models and execute prompts
Read specified project directories or model files for project recommendations and verification
Execute allowlisted LLM Checker CLI commands

Risks and side effects

ollama_pull downloads models and consumes network bandwidth and disk space.
ollama_remove deletes local models and the data they occupy.
ollama_run and cli_exec can run local model or CLI workflows, so prompts and arguments should be reviewed.
Structural acceptance only indicates that a file is structurally safe to load; it does not establish model behavior, provenance, or absence of poisoned weights.
Files above 3 GiB may exceed the WASM32 verification memory ceiling and produce no verdict.

Setup

Before you start

Runtime：Node.js 18+

Install Node.js 18+ and Ollama, then run npm install -g llm-checker. Use llm-checker mcp-setup --client <client> to print or apply client configuration, or add the supplied npx configuration to an MCP-compatible client. Restart the client after changing its configuration.

.mcp.json

{"mcpServers":{"llm-checker":{"command":"npx","args":["--yes","--package","llm-checker","llm-checker-mcp"]}}}

Shown for Claude Code. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "llm-checker": {
      "command": "npx",
      "args": [
        "--yes",
        "--package",
        "llm-checker",
        "llm-checker-mcp"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add llm-checker -- npx --yes --package llm-checker llm-checker-mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

After connecting the client, the tool list should include hw_detect, recommend, ollama_list, etc.; asking the assistant to run hw_detect and getting your actual CPU/GPU/RAM details confirms the server is working.

Troubleshooting

Confirm Node.js 18 or newer and verify that Ollama is installed.
If database commands report missing sql.js, reinstall with npm install -g llm-checker --include=optional.
Run llm-checker hw-detect to inspect hardware detection, then run sync to refresh the Ollama catalog.
Verify that the MCP configuration uses npx --yes --package llm-checker llm-checker-mcp, then restart the client.
For model failures, check Ollama availability, model installation, available memory, and disk space.

Things to try

Once connected, you can ask your AI assistant things like:

What's the best coding model for my hardware?
Compare qwen2.5-coder vs codellama on the same prompt and show tokens/sec
Clean up my Ollama — which installed models should I remove?
Generate optimal Ollama env vars like NUM_GPU and NUM_PARALLEL for my machine

Tools 28

hw_detect read-only

Detect CPU, GPU, memory, and acceleration backends.

check read-only

Run full hardware compatibility analysis and rank models.

recommend read-only

Recommend models by category.

installed read-only

Rank downloaded Ollama models by compatibility.

search read-only

Search the Ollama model catalog with filters.

smart_recommend read-only

Generate advanced recommendations using the full scoring engine.

gpu_plan read-only

Plan single- or multi-GPU model placement and runtime settings.

verify_context read-only

Check a local model's practical context limit against available memory.

Show 20 more tools

ollama_plan read-only

Compute safe Ollama capacity and runtime settings for local models.

ollama_plan_env read-only

Return ready-to-paste Ollama environment variables.

policy_validate read-only

Validate a policy file against the v1 schema.

audit_export writes

Export policy compliance reports for check or recommend flows.

calibrate writes

Generate calibration results and routing policies from a prompt suite.

verify_model read-only

Perform structural safety validation of GGUF or safetensors files with ModelVet.

amd_guard read-only

Run AMD/Windows reliability checks and provide mitigation guidance.

toolcheck read-only

Test tool-calling compatibility for local Ollama models.

ollama_list read-only

List downloaded models with parameters, quantization, family, and size.

ollama_pull writes

Download a model from the Ollama registry.

ollama_run read-only

Run a prompt against a local model and show tokens/sec metrics.

ollama_remove destructive

Delete an Ollama model to free disk space.

ollama_optimize read-only

Generate Ollama optimization environment variables for the hardware.

benchmark read-only

Benchmark a model with three standard prompts and measure runtime metrics.

compare_models read-only

Compare two models side by side on the same prompt.

cleanup_models read-only

Analyze installed models for redundancies, oversized models, and cloud-only models.

project_recommend read-only

Scan a project and recommend a model for its codebase.

ollama_monitor read-only

Show system status, loaded models, and memory headroom.

cli_help read-only

List CLI commands exposed through MCP.

cli_exec read-only

Execute allowlisted LLM Checker CLI commands.

Use cases

Select coding, reasoning, general, or multimodal models for local hardware

Manage, run, benchmark, and compare local Ollama models

Search model artifacts from Hugging Face, Ollama, and GPT4All

Generate hardware-aware Ollama runtime settings

Validate GGUF or safetensors structure before loading a model

Supported clients

Claude Code

Codex

Grok

Kimi Code

Cursor

Windsurf

Gemini CLI

Listed from the project's documentation, not tested by this site.

Overview

LLM Checker is a CLI tool with a built-in MCP server. It detects CPU, GPU, memory, and acceleration backends; scores models deterministically across quality, speed, fit, and context; and recommends models for local inference. Through Ollama it can search, download, run, and remove models. It also supports a multi-source model registry, benchmarking, policy auditing, calibrated routing, project recommendations, and structural safety validation for GGUF and safetensors files.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →

NPM Sentinel MCP Server 79 · B

AI-powered NPM package analysis MCP server

★ 18 · Tools 19 Compare with this →

MartinLoop MCP Server 79 · B

An independent control layer that gives AI coding agents budgets, brakes, and receipts

★ 413 Compare with this →

GitHub MCP Server 76 · B

GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language

★ 33.7k · Tools 15 Compare with this →

Compare with other MCP servers

Source revision 2aa176be96d7 Data synced 2026-10-19 Read the FMRS scoring method