skill-mcp-server

skill-mcp-server is a cursor rule for Cursor from jketreno/clare. It costs 8 tokens per session (2,860 once invoked), scanned A, original, MIT.

A guide for scaffolding an MCP server that exposes CLARE checks as callable tools. MCP, or Model Context Protocol, is a way for AI agents to use external tools through a standard interface.

In plain words
What is it for?
It creates a runnable server with tools for running CI checks and checking the allowed autonomy level for a file path.
Why use it?
It lets agents and automated workflows use project verification and file-permission checks without directly running shell commands or relying on prompt instructions.

Cursor rule for Cursor

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/jketreno/clare/skill-mcp-server
Clone the repo
git clone --depth 1 https://github.com/jketreno/clare

Made for: Cursor.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for skill-mcp-server

README.md
[![agentmods](https://agentmods.dev/badge/rules/jketreno/clare/skill-mcp-server.svg)](https://agentmods.dev/rules/jketreno/clare/skill-mcp-server)
Your own site
<a href="https://agentmods.dev/rules/jketreno/clare/skill-mcp-server"><img src="https://agentmods.dev/badge/rules/jketreno/clare/skill-mcp-server.svg" alt="Measured on agentmods" height="20"></a>
Per session 8 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,860 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00008 $0.02860
Opus 5 $0.00004 $0.01430
Sonnet 5 $0.00002 $0.00572
Haiku 4.5 $0.00001 $0.00286

Measured 4d ago against content hash 31922b0fbbd1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

skill-mcp-server scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

import { execSync } from 'child_process';
.cursor/rules/skill-mcp-server.mdc · 368 lines

How it starts

The opening of the file, as written. The whole thing — 368 lines — stays where its author put it; the contents beside it link to each section on GitHub.


name: mcp-server description: "Scaffold an MCP server exposing CLARE verify and autonomy-check tools" mode: agent

CLARE MCP Server

What this skill does: Scaffolds a minimal MCP (Model Context Protocol) server that exposes CLARE's enforcement primitives — verify-ci.sh and autonomy.yml — as typed tool calls. Any MCP-compatible agent or orchestrator can then call CLARE tools without needing bash access or CLARE-specific prompt engineering.

When to use: When running multi-agent pipelines, headless agents, or any workflow where you need CLARE enforcement available as a structured tool rather than a bash script.

Output: A mcp/ directory containing a runnable MCP server + registration instructions.


Context

CLARE's two core enforcement primitives are:

  1. clare/verify-ci.sh — runs all CI checks, exits non-zero on failure
  2. clare/autonomy.yml — YAML file mapping file paths to autonomy levels

This skill exposes them as three MCP tools:

Tool Input Output
clare_verify (none) {status, passed[], failed[{check, output}], summary}
clare_check_autonomy {path: string} {path, matched_rule, level, reason}
clare_list_humans_only (none) {humans_only_paths: string[]}

Instructions

When this skill is invoked, generate a CLARE MCP server for the current project.

Step 1: Detect the project runtime

Check for package.json → generate Node.js server. Check for pyproject.toml or requirements.txt → generate Python server. If both exist, ask the user which runtime to use. If neither, default to Node.js.

Step 2: Scaffold the server

For Node.js — create mcp/clare-server.js:

#!/usr/bin/env node
// @generated — regenerate from clare/templates/skills/mcp-server.md, do not hand-edit
//
// CLARE MCP Server
// Exposes CLARE enforcement primitives as MCP tool calls.
// See docs/agentic.md for usage in multi-agent pipelines.

import { Server } from '@modelcontextprotocol/sdk/server/index.js';
import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js';
import { CallToolRequestSchema, ListToolsRequestSchema } from '@modelcontextprotocol/sdk/types.js';
import { execSync } from 'child_process';
import { readFileSync } from 'fs';
import { resolve, dirname } from 'path';
import { fileURLToPath } from 'url';
import yaml from 'js-yaml';

const __dirname = dirname(fileURLToPath(import.meta.url));
const PROJECT_ROOT = resolve(__dirname, '..');

const server = new Server(
  { name: 'clare', version: '1.0.0' },
  { capabilities: { tools: {} } }
);

server.setRequestHandler(ListToolsRequestSchema, async () => ({
  tools: [
    {
      name: 'clare_verify',
      description: 'Run clare/verify-ci.sh and return structured pass/fail results. Call this after any code generation before reporting work complete.',
      inputSchema: { type: 'object', properties: {}, required: [] }
    },
    {
      name: 'clare_check_autonomy',
      description: 'Look up the autonomy level for a file path in clare/autonomy.yml. Call this before modifying any file.',
      inputSchema: {
        type: 'object',
        properties: {
          path: { type: 'string', description: 'File path relative to project root' }
        },
        required: ['path']
      }
    },
    {
      name: 'clare_list_humans_only',
      description: 'List all humans-only paths from clare/autonomy.yml. Call this as a pre-flight check before delegating tasks to sub-agents.',
      inputSchema: { type: 'object', properties: {}, required: [] }
    }
  ]
}));

server.setRequestHandler(CallToolRequestSchema, async (request) => {
  const { name, arguments: args } = request.params;

  if (name === 'clare_verify') {
    try {
      const output = execSync(`${PROJECT_ROOT}/clare/verify-ci.sh`, {
        cwd: PROJECT_ROOT,
        encoding: 'utf8',
        stdio: ['pipe', 'pipe', 'pipe']
      });
      return {
        content: [{ type: 'text', text: JSON.stringify({ status: 'passed', output, summary: 'All checks passed' }) }]
      };
    } catch (err) {
      const output = err.stdout + err.stderr;
      const failedChecks = (output.match(/❌ (.+)/g) || []).map(l => l.replace('❌ ', '').trim());
      return {
        content: [{ type: 'text', text: JSON.stringify({ status: 'failed', failed: failedChecks, output, summary: `${failedChecks.length} check(s) failed` }) }]
      };
    }
  }

  if (name === 'clare_check_autonomy') {
    const targetPath = args.path;
    const autonomyFile = readFileSync(`${PROJECT_ROOT}/clare/autonomy.yml`, 'utf8');
    const autonomy = yaml.load(autonomyFile);
    
    const modules = autonomy.modules || [];
    let matched = modules.find(m => m.path !== '*' && targetPath.startsWith(m.path));
    if (!matched) matched = modules.find(m => m.path === '*');
    
    return {
      content: [{
        type: 'text',
        text: JSON.stringify({
          path: targetPath,
          matched_rule: matched?.path || 'none',
          level: matched?.level || 'unknown',
          reason: matched?.reason || ''
        })
      }]
    };
  }

  if (name === 'clare_list_humans_only') {
    const autonomyFile = readFileSync(`${PROJECT_ROOT}/clare/autonomy.yml`, 'utf8');
    const autonomy = yaml.load(autonomyFile);
    const humansOnly = (autonomy.modules || [])
      .filter(m => m.level === 'humans-only')
      .map(m => m.path);
    return {
      content: [{ type: 'text', text: JSON.stringify({ humans_only_paths: humansOnly }) }]
    };
  }

  throw new Error(`Unknown tool: ${name}`);
});

const transport = new StdioServerTransport();
await server.connect(transport);

Read the full file on GitHub · 368 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 368 lines · 8 tokens per session scan A 31922b0fbbd1

Subscribe to this mod's changes

skill-mcp-server is a cursor rule published in the GitHub repository jketreno/clare (5 stars, last pushed 1mo ago), licensed MIT. It adds 8 tokens to every session and 2,860 once invoked, about $0.0000 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.