
cc-configuring-mcp
by jasonkuhrt
Tool configurations
SKILL.md
name: cc-configuring-mcp description: Use when configuring MCP servers in Claude Code, optimizing MCP token usage, troubleshooting MCP connections, understanding Tool Search thresholds, or deciding between MCP tools vs skills for token efficiency
Configuring MCP
Overview
MCP (Model Context Protocol) connects Claude Code to external tools. Core trade-off: MCP tools load at session start (consuming tokens) while skills load on-demand.
Tool Search: Progressive Disclosure for MCP
How It Works
When MCP tool definitions exceed a threshold, Claude Code defers them and uses MCPSearch to discover tools on-demand.
Without Tool Search: 77K tokens consumed before any work
With Tool Search: 8.7K tokens (85% reduction)
Configuration
| Value | Behavior |
|---|---|
auto | Activates at 10% of context (default) |
auto:N | Activates at N% (e.g., auto:5 for 5%) |
true | Always use Tool Search |
false | Never use Tool Search |
Set via environment:
export ENABLE_TOOL_SEARCH=true
Or in settings.json:
{
"env": {
"ENABLE_TOOL_SEARCH": "true"
}
}
When to Use Each Setting
| Scenario | Recommended |
|---|---|
| Use MCP tools every session | auto (default) |
| Use MCP tools occasionally | true or auto:1 |
| Many MCP servers (50+ tools) | auto (triggers automatically) |
| Few MCP servers, want to save tokens | true |
Per-Tool Control: Not Available
Claude API supports per-tool defer_loading:
// API only - NOT in Claude Code
{
default_config: { defer_loading: true },
configs: { "frequently_used_tool": { defer_loading: false } }
}
Claude Code only has global threshold control. Workaround: use rules/instructions to encourage specific MCP tool usage.
Tracking: #12836 (102 upvotes, open)
MCP vs Skills: Token Comparison
| Aspect | MCP (under threshold) | MCP (Tool Search) | Skill |
|---|---|---|---|
| Load timing | Session start | On-demand via search | On-demand via invoke |
| Base cost | Full tool definitions | ~500 tokens (MCPSearch) | 0 tokens |
| Use cost | 0 (already loaded) | Tool definitions loaded | Skill content loaded |
| Best for | Frequent use | Occasional use | Rare use |
Rule of thumb: If you don't use an MCP tool most sessions, either enable Tool Search or use a skill wrapper.
Configuration Files
.mcp.json (Tool Definitions)
{
"mcpServers": {
"server-name": {
"command": "npx",
"args": ["-y", "package-name"],
"env": {
"API_TOKEN": "${API_TOKEN}"
}
}
}
}
settings.json (Enabling Servers)
{
"enableAllProjectMcpServers": true,
"enabledMcpjsonServers": ["server-name"]
}
Note: Servers in .mcp.json but not in enabledMcpjsonServers won't load (0 tokens).
Token-Efficient MCP Design
Action Dispatch Pattern
Instead of many tools, use one tool with actions:
// Bad: 23 tools × ~700 tokens each = 16,100 tokens
"linear_create_issue", "linear_update_issue", "linear_search"...
// Good: 1 tool × ~500 tokens
{"action": "create", ...}, {"action": "update", ...}, {"action": "search", ...}
Example: streamlinear (500 tokens vs 17K for official Linear MCP)
Skill Wrappers for Infrequent MCP
For rarely-used MCP servers:
- Define in
.mcp.jsonbut don't enable by default - Create skill documenting setup + usage
- First use: skill instructs to enable + restart
Troubleshooting
MCP Server Not Loading
- Check
enabledMcpjsonServersincludes server name - Verify env vars are set (use
${VAR}syntax) - Test command manually:
npx -y package-name
Tool Search Not Activating
Check total MCP token usage vs threshold:
- Default 10% of 200K context = 20K tokens
- If under threshold, set
ENABLE_TOOL_SEARCH=trueto force
Model Compatibility
| Model | Tool Search Support |
|---|---|
| Opus 4+ | Yes |
| Sonnet 4+ | Yes |
| Haiku | No (use false) |
Reference
- Claude Code MCP Docs
- Tool Search Announcement
- GitHub #12836 - Per-tool defer_loading request
- GitHub #18303 - Configurable threshold request
スコア
総合スコア
リポジトリの品質指標に基づく評価
SKILL.mdファイルが含まれている
ライセンスが設定されている
100文字以上の説明がある
GitHub Stars 100以上
1ヶ月以内に更新
10回以上フォークされている
オープンIssueが50未満
プログラミング言語が設定されている
1つ以上のタグが設定されている
レビュー
レビュー機能は近日公開予定です



