skills.rest
gujiachun-rainbow
gujiachun-rainbowCOMMUNITY
Explore all skills by this author →
GitHub

tool-creator

Create or upgrade callable Python tools using the @tool decorator.

8|1|Updated Apr 15, 2026
Search Tags:#automation#testing#documentation#python#sandbox#tool-creation
One-click install
npx skills add https://github.com/gujiachun-rainbow/RainClaw --skill tool-creator-gujiachun-rainbow
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: tool-creator
Source: https://github.com/gujiachun-rainbow/RainClaw/tree/main/rainclaw/backend/builtin_skills/tool-creator
Command: npx skills add https://github.com/gujiachun-rainbow/RainClaw --skill tool-creator-gujiachun-rainbow

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill provides a structured, repeatable process to create new executable tools for the agent or upgrade existing ones, enabling more capable and modular automation.

Core Features & Use Cases

  • Defines a standard @tool pattern (one tool per file, named after the function, with full type hints and logging).
  • Guides testing in a sandbox, validating behavior with example tests, and ensuring serializable returns.
  • Supports the "tool creation" workflow and the "tool upgrading" workflow, including documentation and saving via the propose_tool_save flow.

Quick Start

Provide a new tool by detailing its name, inputs, outputs, and behavior, then implement it as an @tool function and test in the sandbox.

Frequently Asked Questions about tool-creator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a Python tool using the @tool decorator?▼

To create a Python tool, implement a function using the @tool decorator with full type hints, logging, and frontmatter-driven docstrings. Maintain a one-tool-per-file structure named after the function for reuse and modularity.

How do I test agent tools in a sandbox environment?▼

Test agent tools in a sandbox by validating behavior with example tests and ensuring serializable returns. Safe sandbox execution allows you to verify callable Python tools before integrating them into the broader workflow.

What is the best way to structure and save new automation tools?▼

The best way to structure new automation tools is enforcing a one-tool-per-file pattern. Save them using the propose_tool_save flow, which ensures strict formatting, explicit documentation, and standardized reuse.

Can I upgrade existing agent tools without breaking the workflow?▼

Yes, you can upgrade existing agent tools by modifying the callable Python function while maintaining strict formatting, type hints, and logging. Test the upgraded tool in the sandbox before saving to ensure safe execution.

Do I need explicit docstrings and type hints for agent tool creation?▼

Yes, explicit frontmatter-driven docstrings and full type hints are required for agent tool creation. This strict formatting ensures the automation tools are callable, serializable, and maintainable across workflows.

Related Skills

Explore More Software Engineering Skills→
open-haxopen-hax
6

opencode-tool-authoring

Define and validate OpenCode tool schemas with YAML frontmatter.

Official
Intermediate
AgentTeam-TaichuAIAgentTeam-TaichuAI
630

tool-creator

Create and upgrade executable Python tools decorated with @tool.

Community
Advanced
ovachieverovachiever
52

grpo-rl-training

Apply GRPO/RL training with TRL to optimize language models.

Community
Advanced
jcl2018jcl2018

CJ_goal_task

Convert small ad-hoc tasks into pull requests via bash scripting.

Community
Intermediate