Agent Spawner

Agent Spawner Skill

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.

Usage

This skill includes a CLI script spawn_agent.py that utilizes agent_utilities.create_agent to build the agent and agent_utilities.chat to send it a prompt and wait for the response.

To use the tool, execute the script with the required arguments.

Examples

Spawning an Agent with a single Prompt:

python -m universal_skills.skills.agent-spawner.scripts.spawn_agent \
    --mcp-config ./references/example_mcp_config.json \
    --prompt "What tools are available to you?"

Spawning an Agent with a specific Name and System Prompt:

python -m universal_skills.skills.agent-spawner.scripts.spawn_agent \
    --mcp-config ./references/example_mcp_config.json \
    --prompt "Please execute a scan" \
    --name "SecurityScanner" \
    --system-prompt "You are a security scanning agent. Always use available tools."

Spawning an Agent using an MCP reference from mcp-client:

# This will automatically resolve 'servicenow-api.json'
# from universal_skills/skills/mcp-client/references/
python -m universal_skills.skills.agent-spawner.scripts.spawn_agent \
    --mcp-config servicenow-api.json \
    --prompt "What tools are available to you?"

Loading configuration from a .env file:

python -m universal_skills.skills.agent-spawner.scripts.spawn_agent \
    --dotenv .env \
    --mcp-config servicenow-api.json \
    --prompt "What tools are available to you?"

Available MCP Servers

You can easily connect to any of these pre-configured MCP servers by passing their JSON filename to the --mcp-config argument.

| MCP Server | Config File | Description | |------------|-------------|-------------| | ServiceNow | servicenow-api.json | ITSM and Workflow automation | | GitLab | gitlab-api.json | DevOps platform for source control & CI/CD | | Microsoft 365 | microsoft-agent.json | Office 365, Outlook, OneDrive, Teams | | Systems Manager | systems-manager.json | System administration and OS-level tools | | Ansible Tower | ansible-tower-mcp.json | Automation controller for Ansible | | Portainer | portainer-agent.json | Container management for Docker/K8s | | Jellyfin | jellyfin-mcp.json | Media server management | | Arr Stack | arr-mcp.json | Media management for PVRs (Sonarr, Radarr) | | Container Manager | container-manager-mcp.json | Docker and container management | | Nextcloud | nextcloud-agent.json | File sharing and collaboration server | | Mealie | mealie-mcp.json | Recipe manager and meal planner | | Repository Manager | repository-manager.json | Maintenance and management of git repos | | Stirling PDF Agent | stirlingpdf-agent.json | PDF manipulation and processing | | Wger Fitness | wger-agent.json | Fitness, workout, and diet tracking |

Arguments

• --prompt: (Required) The instruction/prompt to send to the newly spawned agent.
• --mcp-config: Path to the mcp_config.json file configuring the capabilities. (Defaults to MCP_CONFIG env). If the file doesn't exist locally, it will automatically check for a matching reference in the mcp-client skill.
• --mcp-url: Alternative to config, a URL directly to a single MCP server. (Defaults to MCP_URL env).
• --custom-skills-directory: Path to directory to load custom Universal Skills. (Defaults to CUSTOM_SKILLS_DIRECTORY env).
• --dotenv: Path to a .env file to load environment variables from.
• --name: The name of the spawned agent. (Defaults to DEFAULT_AGENT_NAME env).
• --system-prompt: The system prompt to apply. (Defaults to AGENT_SYSTEM_PROMPT env).
• --provider: The LLM Provider to use inside the agent (e.g. openai, anthropic). (Defaults to PROVIDER env).
• --model-id: The model identifier. (Defaults to MODEL_ID env).
• --base-url: LLM Base URL for API calls. (Defaults to LLM_BASE_URL env).
• --api-key: API key for the LLM. (Defaults to LLM_API_KEY env).
• --insecure: Disable SSL verification for MCP and LLM calls.

Implementation Details

The script depends on agent_utilities which must be installed in the environment. It loads the MCP servers defined in the config file, instantiates a Pydantic AI agent with those MCP servers, and uses the chat method to dispatch the user's prompt to the newly formed agent.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Knuckles-Team
• Source: Knuckles-Team/universal-skills
• License: MIT

Install and usage instructions live in the source repository linked above.