setting-up-mcps
Skill3 starsChanged 3 months agoMITView on GitHub
Guide users through connecting Claude Code to external tools via MCP (Model Context Protocol). Use when setting up new MCP servers, troubleshooting MCP connections, adding integrations like GitHub, Notion, Sentry, or databases, or when user mentions "MCP", "integrations", "connect to tools", or asks how to give Claude access to external services.
- Reads credentials
In this repository14 other files it gives its agents.
- anti-slopSkill
- anti-spamSkill
- careful-marketerSkill
- cro-auditSkill
- keyword-researchSkill
- landing-page-designerSkill
- linkedin-profile-scraperSkill
- listicle-blog-post-writerSkill
- seo-content-strategySkill
- short-form-video-scriptsSkill
- skill-auto-improveSkill
- skill-creatorSkill
- social-media-managerSkill
- web-designerSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 25
Follow this checklist for any new MCP:
- Very specific: line 39
Use --transport http or --transport sse
- Very specific: line 90
Important: All flags (--transport, --env, --scope, --header) must come before the server name.
- Distinctive: line 110
Ask Claude something that uses the tool:
- Very specific: line 147
Windows "Connection closed": Add cmd /c before npx commands
- Very specific: line 149
Auth failing: Use /mcp → "Clear authentication" and retry
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Setting Up MCPs
MCP (Model Context Protocol) connects Claude Code to external tools like GitHub, Notion, databases, and more.
Quick Reference
| Action | Command |
| Add cloud service | claude mcp add --transport http <name> <url> |
| Add local tool | claude mcp add --transport stdio <name> -- <command> |
| List connections | claude mcp list |
| Check status | /mcp (inside Claude Code) |
| Authenticate | /mcp → select service → "Authenticate" |
| Remove connection | claude mcp remove <name> |
Setup Workflow
Follow this checklist for any new MCP:
- [ ] Step 1: Identify the service type (cloud or local)
- [ ] Step 2: Get the MCP URL or command
- [ ] Step 3: Run the add command
- [ ] Step 4: Verify with `claude mcp list`
- [ ] Step 5: Authenticate if needed (`/mcp`)
- [ ] Step 6: Test with a simple requestStep 1: Identify Service Type
Cloud services (GitHub, Notion, Sentry, etc.):
- Use
--transport httpor--transport sse - Require a URL like
https://mcp.servicename.com/... - Usually need browser authentication
Local tools (databases, file servers, custom scripts):
- Use
--transport stdio - Require a command like
npx -y package-name - May need API keys via
--env
Step 2: Get Setup Information
Find the service's MCP documentation. Look for:
- URL for cloud services
- npm package or command for local tools
- Required environment variables (API keys)
Popular hosted MCP servers (check each vendor's own docs for the current endpoint, since these move):
| Service | Transport | Auth |
| GitHub | http | OAuth or a personal access token |
| Notion | http | OAuth |
| Linear | http | OAuth |
| Sentry | http | OAuth |
| Stripe | http | API key header |
| Postgres / SQLite | stdio | connection string via --env |
Vendors change endpoints and auth methods often. Always confirm the field names against the vendor's own MCP documentation rather than copying a config from a third-party aggregator. Different clients expect different keys for the same thing (url vs serverUrl vs httpUrl), and a wrong key fails silently instead of erroring.
Step 3: Run the Add Command
Cloud service (HTTP):
claude mcp add --transport http <name> <url>Cloud service with auth header:
claude mcp add --transport http <name> <url> --header "Authorization: Bearer <token>"Local tool (stdio):
claude mcp add --transport stdio <name> -- <command>Local tool with API key:
claude mcp add --transport stdio <name> --env API_KEY=<value> -- <command>Important: All flags (--transport, --env, --scope, --header) must come before the server name. Use -- to separate the name from the command.
Step 4: Verify Connection
claude mcp listConfirm the new connection appears in the list.
Step 5: Authenticate (If Required)
Inside Claude Code:
- Type
/mcp - Select the service
- Choose "Authenticate"
- Complete browser login
Step 6: Test
Ask Claude something that uses the tool:
- GitHub: "List my open pull requests"
- Notion: "Show my recent pages"
- Database: "List the tables"
Scopes
| Scope | Flag | Use Case |
| local (default) | none | Personal, this project only |
| project | --scope project | Team-shared via .mcp.json in Git |
| user | --scope user | Personal, all your projects |
Example:
claude mcp add --transport http github --scope user https://api.githubcopilot.com/mcp/Management Commands
# List all connections
claude mcp list
# Get details for one
claude mcp get <name>
# Remove connection
claude mcp remove <name>
# Check status inside Claude Code
/mcpTroubleshooting
Quick fixes:
- Windows "Connection closed": Add
cmd /cbeforenpxcommands - Tools not appearing: Restart Claude Code after adding MCPs
- Auth failing: Use
/mcp→ "Clear authentication" and retry - Timeout errors: Set
MCP_TIMEOUT=30000 claudefor longer startup
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

