Instruction file imported from louspringer/OpenFlow-Playground (.cursor/rules/github-mcp-limitations.mdc). Copyright stays with the author.

description: Clarify when to use GitHub MCP servers vs direct tools for local development operations globs: .py,.md,.yaml,.yml,/*.py,/.md,**/.yaml,**/*.yml alwaysApply: false

GitHub MCP Server Limitations

Core Principle

GitHub MCP servers are for repository metadata and operations, NOT for local git operations or GitHub CLI commands. Use direct tools for local development tasks.

What GitHub MCP Servers CAN Do

Repository Operations

• Repository metadata: Get repo info, branches, tags
• Issue tracking: Create, read, update issues
• Pull request management: Create, read, update PRs
• Code review: Comment on code, approve changes
• Release management: Create and manage releases
• Workflow runs: Monitor GitHub Actions

API Operations

• GitHub API calls: Authenticated API requests
• Webhook management: Configure repository webhooks
• Repository settings: Manage repository configuration
• Collaborator management: Add/remove team members

What GitHub MCP Servers CANNOT Do

Local Development Tasks

• Git operations: git add, git commit, git push
• GitHub CLI: gh commands for local operations
• File system operations: Modify local files
• Shell commands: Execute local scripts
• Local authentication: Manage local git credentials

Local Git Workflows

• Branch management: Create/switch local branches
• Staging changes: Add files to git index
• Commit creation: Create local commits
• Push/pull operations: Sync with remote
• Merge operations: Local branch merging

When to Use MCP vs Direct Tools

Use MCP For:

# Repository metadata
repo_info = mcp.get_repository_info("owner/repo")

# Issue creation
issue = mcp.create_issue(
    title="Bug report",
    body="Description of the issue",
    labels=["bug", "help-wanted"]
)

# PR management
pr = mcp.create_pull_request(
    title="Feature implementation",
    body="Description of changes",
    head="feature-branch",
    base="main"
)

Use Direct Tools For:

# Local git operations
import subprocess

def git_commit(message: str):
    """Local git commit"""
    cmd = ["git", "commit", "-m", message]
    subprocess.run(cmd, check=True)

def git_push():
    """Local git push"""
    cmd = ["git", "push"]
    subprocess.run(cmd, check=True)

def create_local_pr():
    """Create PR using local gh CLI"""
    cmd = [
        "gh", "pr", "create",
        "--title", "My PR",
        "--body", "Description",
        "--base", "develop"
    ]
    subprocess.run(cmd, check=True)

Common Misconceptions

❌ Wrong Assumptions

• "MCP can run git commands locally"
• "MCP can execute shell scripts"
• "MCP can modify local files"
• "MCP can handle local authentication"
• "MCP can replace local development tools"

Best Practices

1. Tool Selection

• MCP: Remote repository operations, API calls
• Direct tools: Local git operations, file modifications
• Hybrid: Combine both for complete workflows

2. Error Handling

• Always provide fallbacks when MCP fails
• Use direct tools for critical local operations
• Validate results from both approaches

3. Authentication

• MCP: Uses GitHub API tokens
• Direct tools: Use local git credentials or gh auth
• Keep both authenticated for seamless operation

Remember

GitHub MCP servers are powerful for repository operations but cannot replace local development tools. Use the right tool for each job: MCP for remote, direct tools for local.