Add MCP server configuration to prevent rate limits and context overflow

[Copilot]

MCP code review server was hitting API rate limits and context window limits due to unfiltered file processing.

Changes

Configuration

• .mcp/config.json - Rate limiting (30 req/min, 3 concurrent), context chunking (100k tokens, 10k chunks), file size cap (1MB), caching (1hr TTL)
• .mcpignore - Excludes 156 files (binaries, build artifacts, test coverage, IDE configs)
• .gitignore - Excludes MCP cache directories

Documentation

• MCP_SETUP.md - Setup instructions in repository root
• .mcp/README.md - Full configuration reference
• .mcp/QUICKSTART.md - 5-minute setup for Claude Desktop
• .mcp/IMPACT.md - Performance metrics
• .mcp/claude_desktop_config.example.json - Drop-in example

Impact

Context reduced from 11.58 MB to 0.65 MB (94.4% reduction), 290→134 files. All Python source files retained.

Setup

{
  "mcpServers": {
    "bitbot": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/repo"]
    }
  }
}

The .mcpignore patterns handle the filtering automatically.

Original prompt

My code review mcp servier frequely hita tate limits a nd context limits

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.