tool-scripting

Automate repeated MCP tool calls from the sandbox shell with file-backed logging.

396|51|Updated Mar 14, 2025
One-click install
npx skills add https://github.com/decocms/studio --skill tool-scripting
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tool-scripting
Source: https://github.com/decocms/studio/tree/main/packages/sandbox/image/skills/tool-scripting
Command: npx skills add https://github.com/decocms/studio --skill tool-scripting

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @decocms/typegen.

What problem does it solve?

This Skill removes the friction of repeatedly calling the same MCP tool one item at a time, making bulk and deterministic tool work practical from inside the sandbox.

Core Features & Use Cases

  • Tool Discovery: Inspect the local .deco/tools/ catalog to find available MCP tools and their schemas without loading everything into context.
  • Single or Bulk Execution: Call one tool directly from the shell or automate many calls in a script for rows, files, emails, or other repeated inputs.
  • Scripted Reliability: Use a scratch project, persist results to files, and recover safely from partial failures or expired credentials.
  • Use Case: Process a long list of items through the same org tool and return a concise summary instead of dozens of individual agent calls.

Quick Start

Use the tool-scripting skill to inspect the available tools and run the appropriate MCP tool from the shell or a bulk script.

Frequently Asked Questions about tool-scripting

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

FAQPage Schema
How do I automate batch MCP tool calls from the shell?

Batch MCP tool calls are automated from the shell by scripting repeated executions across many rows, files, or items with deterministic handling and file-backed result logging. The tool-scripting Skill discovers local tool schemas and applies bulk workflows without loading context.

What is the best way to run the same MCP tool across a large list of items?

Running the same MCP tool across a large list of items is best handled through scripted bulk execution from the sandbox shell. This approach processes rows, files, or emails through the same org tool and returns a concise summary instead of individual agent calls.

How does tool schema discovery work for MCP tools in the sandbox?

Tool schema discovery works by inspecting the local `.deco/tools/` catalog to find available MCP tools and their schemas. This allows you to identify endpoint parameters and execute tools directly from the shell without loading everything into context.

Can I recover bulk MCP tool execution from partial failures or expired credentials?

You can recover bulk MCP tool execution from partial failures or expired credentials by using a scratch project and persisting results to files. This scripted reliability ensures deterministic handling and safe recovery during batch automation workflows.

Do I need TypeScript to script shell batch automation for MCP tools?

TypeScript is not required to script shell batch automation for MCP tools. The Skill operates from the sandbox shell to execute bulk workflows across many items, relying on local tool schema discovery and file-backed result logging rather than TypeScript programming.