troubleshoot-mcp-connection
About
This skill diagnoses and resolves MCP server connection issues between Claude Code, Claude Desktop, and MCP servers. It covers Windows argument parsing, authentication failures, transport issues, and platform-specific debugging. Use it when MCP tools don't appear, you encounter "cannot attach the server" errors, or a previously working connection has stopped.
Quick Install
Claude Code
Recommendednpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/troubleshoot-mcp-connectionCopy and paste this command in Claude Code to install this skill
Documentation
Troubleshoot MCP Connection
Diag + resolve MCP server connection failures.
Use When
- Claude Code|Desktop fails to connect to MCP server
- MCP tools don't appear in sess
- "Cannot attach the server" err
- Connection worked → stopped
- Setting up MCP on new machine
In
- Required: Err msg|symptom
- Required: Client (Claude Code, Desktop, both)
- Required: MCP server (mcptools, HF, custom)
- Optional: Recent config|env changes
Do
Step 1: ID Client + Config
Claude Code (WSL):
# View MCP configuration
claude mcp list
claude mcp get server-name
# Configuration stored in
cat ~/.claude.json | python3 -m json.tool
Claude Desktop (Windows):
# Configuration file location
cat "/mnt/c/Users/$USER/AppData/Roaming/Claude/claude_desktop_config.json"
Got: Config file located + readable, MCP server entries w/ command, args, env.
If err: Config doesn't exist|empty → never configured. Follow configure-mcp-server from scratch.
Step 2: Test Server Independently
R mcptools:
# Test if R can start the server
"/mnt/c/Program Files/R/R-4.5.0/bin/Rscript.exe" -e "mcptools::mcp_server()"
Fails:
- R path:
ls "/mnt/c/Program Files/R/" - mcptools installed:
Rscript -e "library(mcptools)" - ellmer dep:
Rscript -e "library(ellmer)"
HF MCP:
# Test mcp-remote directly
mcp-remote https://huggingface.co/mcp
# Check if mcp-remote is installed
which mcp-remote
npm list -g mcp-remote
Got: Server proc starts + init out (JSON-RPC handshake or "listening") no errs.
If err: R mcptools fails → check R ver path correct + mcptools in R lib. mcp-remote fails → reinstall global npm install -g mcp-remote + verify on PATH.
Step 3: Diag Common Err Patterns
"Cannot attach the server" (Claude Desktop)
Root: Windows cmd arg parsing.
Fix: env vars instead of --header:
{
"hf-mcp-server": {
"command": "mcp-remote",
"args": ["https://huggingface.co/mcp"],
"env": { "HF_TOKEN": "your_token" }
}
}
Also ensure mcp-remote global (npm install -g mcp-remote), not relying on npx.
"Connection refused"
- Server not running|wrong port
- Firewall blocking
- Wrong transport (stdio vs HTTP)
"Command not found"
- Missing full path to exec
- PATH not configured in exec ctx
- Windows:
C:\\PROGRA~1\\...for paths w/ spaces
MCP tools missing, no err
- Server starts, tools not registered
- Check server stdout for init msgs
- Verify correct MCP protocol ver
Got: Err pattern matched to documented category (cannot attach, conn refused, cmd not found, silent fail).
If err: No match → capture full err out + check server logs. Search exact err in MCP server's GitHub issues.
Step 4: Network + Auth
# Test Hugging Face API connectivity
curl -I "https://huggingface.co/mcp"
# Verify token validity
curl -H "Authorization: Bearer $HF_TOKEN" https://huggingface.co/api/whoami
Got: HTTP returns 200, whoami returns HF username → confirms net + auth.
If err: curl conn err → DNS|proxy. Token rejected (401) → regen at huggingface.co/settings/tokens + update config.
Step 5: Verify JSON Syntax
# Validate JSON (common issue: trailing commas, missing quotes)
python3 -m json.tool /path/to/config.json
Got: JSON parses no errs → valid syntax.
If err: Common: trailing commas after last entry, missing quotes around strings, mismatched braces. Fix syntax err per parser + re-validate.
Step 6: Platform-Specific Debug
Windows (Claude Desktop):
- Arg parsing differs from Unix
- Spaces in paths break exec
- 8.3 short paths:
C:\PROGRA~1\R\R-45~1.0\bin\x64\Rscript.exe - env vars > cmd-line headers
WSL (Claude Code):
- Unix quoting works
- Full paths w/ spaces (quoted)
- npm/npx via NVM → ensure NVM loaded in exec ctx
Got: Platform-specific issue ID'd (Windows arg parsing, WSL path resolve, NVM ctx).
If err: Windows-specific → switch cmd-line args → env vars for auth. WSL-specific → verify Windows exec path accessible from WSL via full /mnt/c/....
Step 7: Reset + Reconfigure
All else fails:
# Remove and re-add the server (Claude Code)
claude mcp remove server-name
claude mcp add server-name stdio "/full/path/to/executable" -- args
# Restart Claude Desktop after config changes
# (close and reopen the application)
Got: After remove + re-add, claude mcp list shows correct config + fresh conn succeeds.
If err: Re-add fails → check exec path correct + cmd works direct in terminal. Claude Desktop → ensure fully closed (system tray) before restart.
Step 8: Logs
Claude Code: MCP errs in terminal out at sess start.
Claude Desktop: App logs (location varies by OS).
Server-side: Add logging → capture incoming reqs + errs.
Got: Logs reveal specific point of fail (server start, handshake, auth, tool reg).
If err: No logs → add stderr capture to server cmd (redirect to log file) + reproduce. Claude Desktop → %APPDATA%\Claude\logs\ for app logs.
Check
- Server starts independently no errs
- Config JSON valid
- Client connects success
- MCP tools appear in sess
- Tools exec when called
- Conn persists across reqs
Traps
- Wrong config file: Claude Code (
~/.claude.json) vs Desktop (%APPDATA%\Claude\claude_desktop_config.json) - No restart after config: Desktop needs restart; Code picks up on new sess
- npx in restricted env: npx downloads at runtime. Net|perms restricted → install global.
- Token expiration: HF tokens expire. Regen if auth fails suddenly.
- Version mismatch: MCP protocol vers must be compat between client + server
→
configure-mcp-server— initial MCP setupbuild-custom-mcp-server— custom server debug ctxsetup-wsl-dev-environment— WSL prereq setup
GitHub Repository
Related Skills
evaluating-llms-harness
TestingThis Claude Skill runs the lm-evaluation-harness to benchmark LLMs across 60+ standardized academic tasks like MMLU and GSM8K. It's designed for developers to compare model quality, track training progress, or report academic results. The tool supports various backends including HuggingFace and vLLM models.
cloudflare-cron-triggers
TestingThis skill provides comprehensive knowledge for implementing Cloudflare Cron Triggers to schedule Workers using cron expressions. It covers setting up periodic tasks, maintenance jobs, and automated workflows while handling common issues like invalid cron expressions and timezone problems. Developers can use it for configuring scheduled handlers, testing cron triggers, and integrating with Workflows and Green Compute.
webapp-testing
TestingThis Claude Skill provides a Playwright-based toolkit for testing local web applications through Python scripts. It enables frontend verification, UI debugging, screenshot capture, and log viewing while managing server lifecycles. Use it for browser automation tasks but run scripts directly rather than reading their source code to avoid context pollution.
finishing-a-development-branch
TestingThis skill helps developers complete finished work by verifying tests pass and then presenting structured integration options. It guides the workflow for merging, creating PRs, or cleaning up branches after implementation is done. Use it when your code is ready and tested to systematically finalize the development process.
