Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate MCP capability declarations from tool and resource inventory with proper versioning and feature flags.
.claude/skills/a5c-ai-mcp-capability-declarator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-03 | ✗→✗ | = Same ✗ | -23% | 0% |
Generate MCP capability declarations from tool/resource inventory.
Invoke this skill when you need to:
| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | serverName | string | Yes | Server name | | version | string | Yes | Server version | | tools | array | No | Tool capabilities | | resources | array | No | Resource capabilities | | prompts | array | No | Prompt capabilities |
typescriptimport { ServerCapabilities, InitializationOptions } from '@modelcontextprotocol/sdk/types.js'; // Server metadata export const SERVER_INFO = { name: 'my-mcp-server', version: '1.0.0', } as const; // Capability declarations export const CAPABILITIES: ServerCapabilities = { tools: { // Tool capabilities listChanged: true, }, resources: { // Resource capabilities subscribe: true, listChanged: true, }, prompts: { // Prompt capabilities listChanged: true, }, logging: {}, }; // Tool definitions export const TOOL_DEFINITIONS = [ { name: 'search_files', description: 'Search for files matching a pattern', inputSchema: { type: 'object', properties: { pattern: { type: 'string', description: 'Glob pattern' }, path: { type: 'string', description: 'Base path' }, }, required: ['pattern'], }, }, { name: 'read_file', description: 'Read contents of a file', inputSchema: { type: 'object', properties: { path: { type: 'string', description: 'File path' }, }, required: ['path'], }, }, ] as const; // Resource templates export const RESOURCE_TEMPLATES = [ { uriTemplate: 'file:///{path}', name: 'File', description: 'Access file contents', mimeType: 'text/plain', }, ] as const; // Create initialization options export function createInitializationOptions(): InitializationOptions { return { serverInfo: SERVER_INFO, capabilities: CAPABILITIES, }; } // Capability documentation export const CAPABILITY_DOCS = ` # ${SERVER_INFO.name} v${SERVER_INFO.version} ## Supported Capabilities ### Tools ${TOOL_DEFINITIONS.map(t => `- **${t.name}**: ${t.description}`).join('\n')} ### Resources - File access via \`file:///\` URI scheme - Resource subscription support - List change notifications ### Prompts - Dynamic prompt templates - List change notifications ## Feature Flags - Tool list change notifications: enabled - Resource subscriptions: enabled `;
pythonfrom dataclasses import dataclass from typing import List, Dict, Any @dataclass class ServerInfo: name: str = "my-mcp-server" version: str = "1.0.0" @dataclass class ToolDefinition: name: str description: str input_schema: Dict[str, Any] TOOL_DEFINITIONS: List[ToolDefinition] = [ ToolDefinition( name="search_files", description="Search for files matching a pattern", input_schema={ "type": "object", "properties": { "pattern": {"type": "string"}, "path": {"type": "string"}, }, "required": ["pattern"], }, ), ] CAPABILITIES = { "tools": {"listChanged": True}, "resources": {"subscribe": True, "listChanged": True}, "prompts": {"listChanged": True}, "logging": {}, } def create_initialization_options() -> dict: return { "serverInfo": { "name": ServerInfo.name, "version": ServerInfo.version, }, "capabilities": CAPABILITIES, }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 11,852 | 7,609 | -36% | 1 | 1 | 0% | 2,713 | 2,809 | +4% | 0 | 0 | — |
case-02 | fail→pass | 9,613 | 5,941 | -38% | 1 | 1 | 0% | 2,117 | 2,455 | +16% | 0 | 0 | — |
case-03 | fail→fail | 18,012 | 11,074 | -39% | 1 | 1 | 0% | 4,066 | 3,121 | -23% | 0 | 0 | — |
case-04 | fail→fail | 6,236 | 4,664 | -25% | 1 | 1 | 0% | 1,117 | 2,114 | +89% | 0 | 0 | — |
case-05 | fail→fail | 8,488 | 5,124 | -40% | 1 | 1 | 0% | 1,704 | 2,112 | +24% | 0 | 0 | — |
case-06 | fail→fail | 10,715 | 6,592 | -38% | 1 | 1 | 0% | 1,988 | 2,438 | +23% | 0 | 0 | — |
case-07 | fail→fail | 7,223 | 3,673 | -49% | 1 | 1 | 0% | 1,405 | 1,889 | +34% | 0 | 0 | — |
case-08 | fail→fail | 10,404 | 5,214 | -50% | 1 | 1 | 0% | 2,058 | 2,166 | +5% | 0 | 0 | — |
case-09 | fail→fail | 4,262 | 5,386 | +26% | 1 | 1 | 0% | 832 | 2,029 | +144% | 0 | 0 | — |
case-10 | fail→fail | 9,699 | 6,057 | -38% | 1 | 1 | 0% | 1,816 | 2,319 | +28% | 0 | 0 | — |
case-11 | fail→pass | 11,919 | 9,087 | -24% | 1 | 1 | 0% | 1,981 | 3,045 | +54% | 0 | 0 | — |
case-12 | fail→fail | 6,929 | 6,052 | -13% | 1 | 1 | 0% | 1,497 | 2,349 | +57% | 0 | 0 | — |
case-13 | fail→pass | 14,993 | 9,459 | -37% | 1 | 1 | 0% | 2,747 | 2,981 | +9% | 0 | 0 | — |
case-14 | fail→fail | 7,166 | 5,012 | -30% | 1 | 1 | 0% | 1,336 | 2,168 | +62% | 0 | 0 | — |
case-15 | fail→fail | 6,463 | 4,674 | -28% | 1 | 1 | 0% | 1,294 | 2,074 | +60% | 0 | 0 | — |
case-16 | fail→fail | 11,349 | 6,186 | -45% | 1 | 1 | 0% | 2,213 | 2,419 | +9% | 0 | 0 | — |
case-17 | fail→fail | 7,880 | 5,357 | -32% | 1 | 1 | 0% | 1,581 | 2,110 | +33% | 0 | 0 | — |
case-18 | fail→fail | 9,099 | 4,687 | -48% | 1 | 1 | 0% | 1,707 | 2,080 | +22% | 0 | 0 | — |
case-19 | fail→fail | 5,435 | 5,803 | +7% | 1 | 1 | 0% | 1,078 | 2,310 | +114% | 0 | 0 | — |
case-20 | fail→fail | 12,285 | 6,112 | -50% | 1 | 1 | 0% | 2,895 | 2,550 | -12% | 0 | 0 | — |
case-21 | fail→fail | 10,437 | 5,294 | -49% | 1 | 1 | 0% | 1,897 | 2,278 | +20% | 0 | 0 | — |
case-22 | fail→fail | 9,927 | 9,015 | -9% | 1 | 1 | 0% | 2,031 | 2,621 | +29% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.