Tooluniverse Custom Tool

This extends ToolUniverse's 1000+ built-in tools with your own custom ones for APIs or computations it doesn't already cover. You get three options: JSON configs for simple REST endpoints (no code), Python classes in your workspace for local use, or full plugin packages you can share via pip. The JSON approach is genuinely simple, just endpoint and parameter schema. Python classes give you control over parsing and state but have different signatures depending on whether you're building workspace tools (keyword args) or plugins (single dict arg). Worth noting that tu test validates return schemas automatically but passes empty arrays, so use test examples that return real data to catch issues.

skills.sh installs: 290
GitHub stars: 1.8K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add mims-harvard/tooluniverse --skill tooluniverse-custom-tool --agent claude-code

Installs into .claude/skills of the current project.

Files

Adding Custom Tools to ToolUniverse

When to create a custom tool: Create one if you need to access an API that ToolUniverse doesn't cover, or if you need a specialized data transformation that no existing tool provides. Start with the JSON config approach (simplest — no Python needed); escalate to a Python class only if you need custom response parsing or stateful logic.

Three ways to add tools — pick the one that fits your needs:

Approach	When to use
JSON config	REST API with standard request/response — no coding needed
Python class (workspace)	Custom logic for local/private use only
Plugin package	Reusable tools you want to share or install via pip