Skip to content
Back to skills

Generate

BSecurity

Turn a codebase, an OpenAPI/gRPC/GraphQL spec, or a plain-language description of an API into a production-ready MCP server, using mcp-anything. Use when the user wants to expose software, an API, a CLI tool, or a service as MCP tools for Claude or other agents.

  • 48 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 23, 2026
ai-agentstypescriptpythonbashgitapi

Works with

  • cli
  • api
  • mcp

Security analysis

B84/100
  • mediumInstalls packages at runtime which could introduce malicious dependencies
  • mediumInstalls packages at runtime which could introduce malicious dependencies

Pro shows the line behind each finding and how to fix it

Scanned September 23, 2026

npx -y skills add Type-MCP/mcp-anything --skill generate --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Generate?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Generate
[![Security: B — Skills Directory](https://www.skillsdirectory.com/api/skills/type-mcp-generate/badge)](https://www.skillsdirectory.com/skills/type-mcp-generate)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: generate
description: Turn a codebase, an OpenAPI/gRPC/GraphQL spec, or a plain-language description of an API into a production-ready MCP server, using mcp-anything. Use when the user wants to expose software, an API, a CLI tool, or a service as MCP tools for Claude or other agents.
when_to_use: "Use when the user asks to 'turn X into an MCP server', 'make an MCP server for Y', 'wrap this API/CLI as MCP tools', or wants agents to be able to call their software."
argument-hint: "[path-or-url] [--target fastmcp|mcp-use]"
allowed-tools: Bash(mcp-anything *) Bash(pip install *) Bash(pip show *) Bash(command -v mcp-anything)
---

## Generate an MCP server with mcp-anything

mcp-anything (https://github.com/gabrielekarra/mcp-anything) turns a codebase, an API spec, or a
plain-language brief into a working MCP server — Python/FastMCP by default, or TypeScript/mcp-use
with `--target mcp-use`.

### 1. Make sure the CLI is current

The MCP ecosystem moves fast; prefer the latest source over a possibly-stale PyPI release:

!`command -v mcp-anything >/dev/null 2>&1 && mcp-anything --version || echo "not installed"`

If it's missing, install from the repo (not just `pip install mcp-anything` — PyPI can lag):

```bash
pip install --upgrade "mcp-anything[llm] @ git+https://github.com/gabrielekarra/mcp-anything.git"
```

### 2. Find out what to expose

Ask only what's missing:
- **The source**: a local codebase path, or an OpenAPI/GraphQL/gRPC spec (file path or URL).
- **What agents should be able to do with it** — a sentence or two of use cases is enough for
  the richer path below.

### 3. Pick a path

- **Quick scan** (no use-case framing given): `mcp-anything generate <path-or-url> -o <output-dir>`
- **Recommended — curated, agent-optimized tools** (use cases given): write a short domain brief
  to a temp YAML file (`server_name`, `description`, `use_cases: [...]`), then:
  `mcp-anything build --brief <brief.yaml> --data-source <path-or-spec> -o <output-dir> --no-llm`
  — drop `--no-llm` when `ANTHROPIC_API_KEY` is set; grouping and descriptions are noticeably
  better with it.
- Add `--target mcp-use` to either command for TypeScript instead of Python.

### 4. Report back

`mcp-anything` prints the output directory, tool count, and an MCP client config snippet
(`mcpServers: {...}`) — surface that, plus how to try it: `pip install -e <output-dir>` (Python)
or `npm install` (TypeScript) in the output directory, then add the snippet to the user's client
config.

### 5. Optional: validate

`mcp-anything validate <output-dir>` checks the generated server against mcp-anything's own
output contract (tool shape, transport, telemetry, packaging).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…