Ask LLM Claude MCP Server is one of the seven MCP servers in the same repository, with a narrow, clear scope: it gives non-Claude MCP clients a read-only consulting channel to the Claude Code CLI. It exposes only ask-claude, ping, and get-usage-stats, and its manifest declares four optional environment variables covering model, fallback, timeout, and log level. It is simple and restrained in permissions (Read/Glob/Grep), which suits using Claude as an independent reviewer. The limits are equally clear: it requires a separately installed and authenticated Claude Code CLI, consumes that quota, sends data to Claude, and is unofficial at version 0.0.1.
Strongest · Documentation 14/20 Weakest · Reliability 9/20
Static review · not runReviewed 2026-09-25
Read the FMRS scoring method →Runtime：Node.js 20+
ASK_CLAUDE_MODEL
Default Claude model alias or full model name (default: opus); nothing to fetch, change it to pick another model.
ASK_CLAUDE_FALLBACK_MODEL
Claude CLI fallback model used when the default is overloaded or unavailable (default: sonnet).
ASK_CLAUDE_TIMEOUT_MS
Timeout for Claude CLI execution in milliseconds (default: 600000, i.e. 10 minutes).
GMCPT_LOG_LEVEL
Log verbosity: debug, info, warn, error (default: warn).
codex mcp add claude -- npx -y @ask-llm/claude-mcp; for other clients add the STDIO config {"command":"npx","args":["-y","@ask-llm/claude-mcp"]}.{
"mcpServers": {
"claude": {
"command": "npx",
"args": ["-y", "@ask-llm/claude-mcp"],
"env": {
"ASK_CLAUDE_MODEL": "opus",
"ASK_CLAUDE_FALLBACK_MODEL": "sonnet",
"ASK_CLAUDE_TIMEOUT_MS": "600000",
"GMCPT_LOG_LEVEL": "warn"
}
}
}
}
Shown for Claude Code. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.
{
"servers": {
"claude": {
"command": "npx",
"args": [
"-y",
"@ask-llm/claude-mcp"
],
"env": {
"ASK_CLAUDE_MODEL": "opus",
"ASK_CLAUDE_FALLBACK_MODEL": "sonnet",
"ASK_CLAUDE_TIMEOUT_MS": "600000",
"GMCPT_LOG_LEVEL": "warn"
}
}
}
}
Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).
claude mcp add claude -e ASK_CLAUDE_MODEL=opus -e ASK_CLAUDE_FALLBACK_MODEL=sonnet -e ASK_CLAUDE_TIMEOUT_MS=600000 -e GMCPT_LOG_LEVEL=warn -- npx -y @ask-llm/claude-mcp
Run it in a terminal; replace any <…> placeholders with your own values first.
The client's tool list should show ask-claude, get-usage-stats and ping; a prompt like "ask Claude for an independent opinion on this code" that returns a Claude response confirms the connection works.
Once connected, you can ask your AI assistant things like:
No matching tools
Listed from the project's documentation, not tested by this site.
This is the server defined by packages/claude-mcp/server.json in the Ask LLM repository (npm package @ask-llm/claude-mcp). It bridges Codex and other MCP clients with the Anthropic Claude Code CLI so a non-Claude assistant can request read-only second opinions and code reviews. It runs over stdio and exposes the ask-claude, ping, and get-usage-stats tools; ask-claude defaults to the opus model with a sonnet fallback and grants read-only Read/Glob/Grep workspace access. It fits workflows where a non-Claude primary assistant (for example Codex) wants Claude's independent take on a diff, plan, or approach before committing. It is not for cases where Claude must write or modify files, and it is not the unified multi-provider orchestrator from the same repository (@ask-llm/mcp). It requires a separately installed and authenticated Claude Code CLI, and it is an unofficial third-party tool.
Upstash's official server providing up-to-date third-party library docs for AI coding assistants
AI-powered NPM package analysis MCP server
An independent control layer that gives AI coding agents budgets, brakes, and receipts
GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language
Data synced 2026-09-25 Read the FMRS scoring method