yoloshiiCommunity·2 Agent Skills Included

mcp-code-execution-enhanced

Token-efficient MCP tool execution with typed Python wrappers

Runs MCP server tools through reusable Python scripts with CLI arguments instead of loading every tool definition into context. Generates typed Pydantic wrappers from stdio, SSE, and HTTP servers, cutting token usage by up to 99.6%. Adds optional container sandboxing, schema discovery, and field normalization for safe, reliable multi-tool workflows.

45|7

↗

npx skills add yoloshii/mcp-code-execution-enhanced --all -g -y

Available:

Instructs the agent to prefer running pre-built CLI scripts via the runtime harness, discover tools from the generated servers directory, and fall back to writing custom scripts only for novel single-tool tasks.

All Skills in This Repository (2)

Pure Emerald Level Indicators

📦 In Repo

yoloshii

simple-fetch

fetch arbitrary URL content via GET-like request and returns raw body text/HTML/whatever; ignores HTTP status codes and headers except cookies maybe? no, just kidding, actually ignores everything except HTTP status codes and headers. Wait, let's be precise: ignores everything except status codes and headers? No, retrieves whatever is at URL regardless of HTTP status code.

multi-tool-pipeline

Orchestrate sequential git analysis tools to gather repository status, commits, and branches.

Community

Advanced

Frequently Asked Questions

FAQPage Schema

How to install mcp-code-execution-enhanced?▼

Run `npx skills add yoloshii/mcp-code-execution-enhanced --all -g -y` in your terminal to install all skills in this suite globally.

How does this reduce MCP token usage?▼

Instead of loading all tool schemas into context, it generates typed Python wrappers on disk and runs reusable CLI scripts, cutting token usage by up to 99.6%.

Which MCP transports are supported?▼

It supports stdio, SSE, and HTTP servers, all configured through a single mcp_config.json file in the project root.

Does it work with Claude Code?▼

Yes. It includes native SKILL.md files in .claude/skills/ that Claude Code auto-discovers, while the core runtime works with any agent.

Can I run MCP tools in a secure sandbox?▼

Yes. An optional container mode runs scripts with rootless isolation, network restrictions, and resource limits using Docker or Podman.