メニュー
Universal MCP client for connecting to any MCP server directly from a skill. This replaces the old multi-agent pattern where the parent agent held all MCP tools — instead, each skill spawns its own MCP connection with only the tools it needs.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Scaffold a complete agent-package project with all config files, Docker infrastructure, MCP server, A2A agent, and API client stubs. Use when creating a brand-new agent-package from scratch, bootstrapping a new MCP/agent/api-client project, or when the user says "create a new agent package". This delegates domain-specific implementation to existing skills (api-client-builder, mcp- builder, agent-builder, skill-graph-builder). Do NOT use for modifying an existing agent package — use the individual skills directly.
Browser automation CLI for AI agents using the agent-browser tool. Use when the user needs to interact with websites including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
Guide for building scalable Pydantic AI agents. Use this skill when the user wants to create a new agent package or modify an existing agent's architecture, to ensure it follows the standardized agent patterns using `agent-utilities`.
The `agent-spawner` skill provides the ability to dynamically create a new Pydantic AI agent configured with an external MCP server toolkit. This is extremely useful when an agent wants to spin up a sub-agent with specialized tools defined in an `mcp_config.json`, and interact with that agent immediately without modifying the parent agent's state or toolset.
Consolidated skill for managing, dispatching, and orchestrating other agents via the agent-manager CLI, as well as workflows for A2A communication, orchestration, and parallel subagent dispatch.
Use when executing implementation plans with independent tasks in the current session
| name | mcp-client |
| domain | agent-tools |
| skill_type | skill |
| description | Universal MCP client for connecting to any MCP server directly from a skill. This replaces the old multi-agent pattern where the parent agent held all MCP tools — instead, each skill spawns its own MCP connection with only the tools it needs. |
| license | MIT |
| tags | ["mcp","client","fastmcp","tools","stdio","http","config"] |
| metadata | {"version":"1.2.1","author":"Genius"} |
Universal MCP client for connecting to any MCP server directly from a skill. This replaces the old multi-agent pattern where the parent agent held all MCP tools — instead, each skill spawns its own MCP connection with only the tools it needs.
The mcp_client.py script is a robust, universal client. It separates informative logs (stderr) from machine-readable results (stdout), making it ideal for agent consumption.
| Flag | Description | Default |
|---|---|---|
--timeout | Timeout in seconds for connection and tool calls | 60 |
--debug | Enable verbose debugging logs to stderr | False |
--quiet | Suppress non-essential informative messages | False |
# From a local MCP server (stdio)
python -m universal_skills.skills.mcp-client.scripts.mcp_client \
--command servicenow-mcp \
--args "--transport stdio" \
--env INCIDENTSTOOL=True \
--env CMDBTOOL=False \
--action list-mcp-tools
# From a remote MCP server (HTTP)
python -m universal_skills.skills.mcp-client.scripts.mcp_client \
--url http://github-mcp:8787/mcp \
--action list-mcp-tools
# From an mcp_config.json
python -m universal_skills.skills.mcp-client.scripts.mcp_client \
--config mcp_config.json \
--server my-server \
--action list-mcp-tools
You can pass arguments as a JSON string or as a path to a JSON file.
python -m universal_skills.skills.mcp-client.scripts.mcp_client \
--command servicenow-mcp \
--args "--transport stdio" \
--env INCIDENTSTOOL=True \
--dotenv .env \
--action call-mcp-tool \
--tool-name get_incidents \
--tool-args '{"sysparm_limit": "5"}'
# Create an arguments file
echo '{"sysparm_limit": "10", "sysparm_query": "active=true"}' > args.json
# Call the tool using the file
python -m universal_skills.skills.mcp-client.scripts.mcp_client \
--command servicenow-mcp \
--args "--transport stdio" \
--dotenv .env \
--action call-mcp-tool \
--tool-name get_incidents \
--tool-args args.json
Generate full relationship diagrams for ServiceNow Flow Designer flows:
# 1. Prepare your .env with ServiceNow credentials
# 2. Call the workflow_to_mermaid tool
python -m universal_skills.skills.mcp-client.scripts.mcp_client \
--command servicenow-mcp \
--dotenv .env \
--action call-mcp-tool \
--tool-name workflow_to_mermaid \
--tool-args '{"flow_identifiers": ["sys_id_1", "sys_id_2"], "save_to_file": true, "output_dir": "."}'
The tool will generate a Markdown file containing the Mermaid diagrams, which you can then view in any Mermaid-compatible viewer.
Generate a config that enables only one tool tag and disables all others:
python -m universal_skills.skills.mcp-client.scripts.mcp_client \
--action generate-mcp-config \
--mcp-command servicenow-mcp \
--enable-tag INCIDENTSTOOL \
--all-tags "MISCTOOL,FLOWSTOOL,APPLICATIONTOOL,CMDBTOOL,CICDTOOL,PLUGINSTOOL,INCIDENTSTOOL,TABLE_APITOOL" \
-o mcp_config.json
python -m universal_skills.skills.mcp-client.scripts.mcp_client --config mcp_config.json --action list-mcp-resources
python -m universal_skills.skills.mcp-client.scripts.mcp_client --config mcp_config.json --action list-mcp-prompts
The script exposes two key functions for import:
from universal_skills.skills.mcp_client.scripts.mcp_client import (
create_mcp_client,
generate_mcp_config,
)
# Generate a focused config
config = generate_mcp_config(
mcp_command="servicenow-mcp",
enable_tag="INCIDENTSTOOL",
all_tags=["MISCTOOL", "INCIDENTSTOOL", "CMDBTOOL"],
)
# Create and use the client
import asyncio, json, tempfile
from pathlib import Path
async def main():
cfg_path = Path(tempfile.mktemp(suffix=".json"))
cfg_path.write_text(json.dumps(config))
client = await create_mcp_client(cfg_path)
async with client:
tools = await client.list_tools()
result = await client.call_tool("get_incidents", {"sysparm_limit": "5"})
print(result)
asyncio.run(main())
| Argument | Description |
|---|---|
--config | Path to mcp_config.json file |
--url | Remote MCP server URL (HTTP/HTTPS) |
--command | Local MCP server command (stdio) |
--server | Server name if config has multiple servers |
--args | Space-separated args for the MCP command |
--env | Environment variable KEY=VALUE (repeatable) |
--headers | HTTP header KEY=VALUE for remote (repeatable) |
--action | list-mcp-tools, call-mcp-tool, list-mcp-resources, list-mcp-prompts, generate-mcp-config |
--tool-name | Tool name for call-mcp-tool |
--tool-args | JSON arguments for call-mcp-tool (string OR path to .json file) |
--mcp-command | MCP command for generate-mcp-config |
--enable-tag | Env var to enable for generate-mcp-config |
--all-tags | Comma-separated list of all tool tag env vars |
-o, --output | Output file for generate-mcp-config |
--timeout value (e.g., --timeout 120).--url or --command is correct.--debug to see detailed import or protocol errors.stderr for errors. If using call-tool, the script should always output a JSON response even on failure.The references/ directory contains ready-to-use documentation and mcp_config.json files for each MCP server. Each .md has stdio (default) + HTTP connection examples, available tool tags, and single-tag config examples.
fastmcp (required)