text2agent_skill

Overview

This skill automates creation of purpose-built Python agents when you explicitly request a new agent. It analyzes your requirements, synthesizes a robust architecture from a search foundation plus relevant specialist patterns, and generates ready-to-run agent code and configuration files. Use it only when you need a new agent created programmatically.

How this skill works

The skill performs a deep requirements analysis to extract the core objective, identity, capabilities, system prompt tone, and MCP/tool needs. It then fuses a reliable search-based foundation with domain-specific patterns from reference agents to produce a tailored system prompt, tool list, and MCP configuration. Finally it scaffolds the agent directory, writes Python implementation and config files, verifies file creation, registers the agent, and performs dependency checks to ensure MCP servers can launch.

When to use it

• You explicitly request an agent to be created (code + config).
• You need a Python agent with integrated MCP servers and terminal support.
• You want a fused design combining a robust search workflow with a domain specialist’s logic.
• You need an agent registered dynamically and environment-prepared for runtime.
• You require reproducible agent scaffolding saved to the agents directory.

Best practices

• Provide a clear, concise core objective and expected outputs to guide system prompt synthesis.
• List required external tools, APIs, and any privileged access or credentials up front.
• Specify desired personality, tone, and strict workflow constraints if important.
• Include sample inputs and desired output formats to shape the agent’s output logic.
• Limit use to explicit agent-creation requests — this skill is not for general tasks.

Example use cases

• Create a customer-support agent that searches knowledge bases, formats responses, and logs interactions to files.
• Build a data-collection agent that runs web searches, downloads artifacts, and runs local analysis pipelines.
• Scaffold a financial-reporting agent that follows a multi-step workflow: gather data, compute KPIs, add disclaimers, and format results.
• Generate an RL-training orchestrator that coordinates environment servers and training monitors via MCP servers.
• Produce a domain-specific assistant that combines search robustness with specialist guardrails and output templates.