notque/vexjoy-agent / skills/process/workflow
workflow
VerifiedStructured work: multi-phase tasks, feature builds, planning, objective loops, hill climbing.
references/mcp-pipeline-builder/references/python-scaffold-template.md
View on GitHub# Python FastMCP Server Scaffold Template
Reference for Phase 3 GENERATE when `--python` flag is set. Use these patterns for Python MCP servers.
---
## When to Use Python Mode
Generate a Python FastMCP server (instead of TypeScript) when:
1. The user explicitly passes `--python`, AND
2. The target repo is a Python project (has `pyproject.toml`, `setup.py`, or `requirements.txt`)
Python mode is preferred when the target exports an installable package that can be imported directly (`--import` flag). Otherwise, TypeScript is simpler and more type-safe.
---
## Directory Structure
```
{service}-mcp-server/
pyproject.toml
src/
main.py ← @mcp.tool decorators, mcp.run()
client.py ← Shared client class (HTTP or subprocess)
README.md
```
---
## `pyproject.toml` Template
```toml
[project]
name = "{service}-mcp-server"
version = "0.1.0"
description = "MCP server for {Service Name}"
requires-python = ">=3.11"
dependencies = [
"fastmcp>=0.1.0",
"httpx>=0.27.0", # For HTTP client; remove if not needed
"pydantic>=2.0.0", # Included transitively by fastmcp
]
[build-system]
requires = ["hatchling"]
build-backend = "hatchling.build"
[tool.hatch.build.targets.wheel]
packages = ["src"]
```
---
## `src/main.py` Template
```python
"""
{Service Name} MCP server.
Provides tools for {one-line purpose}.
Environment variables:
SERVICE_API_KEY: Required. API key for authenticating with {Service Name}.
"""
import os
import json
from typing import Annotated
import fastmcp
from fastmcp import FastMCP
from pydantic import BaseModel, Field
from .client import ServiceClient
# Validate required env vars at startup
_API_KEY = os.environ.get("SERVICE_API_KEY")
if not _API_KEY:
raise ValueError(
"SERVICE_API_KEY environment variable is required. "
"Set it before starting the server."
)
# Initialize FastMCP
mcp = FastMCP("{service}-mcp-server")
# Initialize shared client
_client = ServiceClient(api_key=_API_KEY)
# --- Tool parameter models (Pydantic v2) ---
class GetIssueParams(BaseModel):
owner: str = Field(description="Repository owner or organization name")
repo: str = Field(description="Repository name")
issue_number: int = Field(description="Issue number", gt=0)
class ListIssuesParams(BaseModel):
owner: str = Field(description="Repository owner or organization name")
repo: str = Field(description="Repository name")
state: str = Field(default="open", description="Filter by state: open, closed, all")
limit: int = Field(default=20, ge=1, le=100, description="Maximum results. Default: 20")
# --- Tools ---
@mcp.tool(
description="Get a single issue by number. Returns title, description, state, labels, and assignees."
)
def service_get_issue(params: GetIssueParams) -> str:
"""Get a single issue with full context."""
try:
issue = _client.get_issue(params.owner, params.repo, params.issue_number)
return json.dumps(issue, indent=2)
except Exception as e:
# Raise with a clear message — FastMCP converts this to an MCP error response
raise ValueError(f"Error fetching issue {params.issue_number}: {e}") from e
@mcp.tool(
description="List issues with optional state filter. Returns a JSON array of matching issues."
)
def service_list_issues(params: ListIssuesParams) -> str:
"""List issues with filters."""
try:
issues = _client.list_issues(params.owner, params.repo, params.state, params.limit)
return json.dumps(issues, indent=2)
except Exception as e:
raise ValueError(f"Error listing issues: {e}") from e
# --- Entry point ---
if __name__ == "__main__":
mcp.run(transport="stdio")
```
---
## `src/client.py` Template
The shared client class. Adapt to the target service's protocol.
### HTTP Client Pattern
```python
"""Shared HTTP client for {Service Name} API."""
import httpx
from typing import Any
class ServiceClient:
def __init__(self, api_key: str, base_url: str = "https://api.example.com"):
self._api_key = api_key
self._base_url = base_url.rstrip("/")
def _get(self, path: str, params: dict[str, str] | None = None) -> Any:
"""Make an authenticated GET request."""
with httpx.Client() as client:
response = client.get(
f"{self._base_url}{path}",
params=params,
headers={
"Authorization": f"Bearer {self._api_key}",
"Accept": "application/json",
},
timeout=30.0,
)
response.raise_for_status()
return response.json()
def get_issue(self, owner: str, repo: str, number: int) -> dict:
return self._get(f"/repos/{owner}/{repo}/issues/{number}")
def list_issues(self, owner: str, repo: str, state: str, limit: int) -> list:
return self._get(
f"/repos/{owner}/{repo}/issues",
params={"state": state, "per_page": str(limit)},
)
```
### Subprocess/CLI Client Pattern
```python
"""Shared subprocess client for {Service Name} CLI."""
import subprocess
import json
import os
class CliClient:
def __init__(self, executable: str, env: dict[str, str] | None = None):
self._executable = executable
self._env = {**os.environ, **(env or {})}
def run(self, args: list[str]) -> str:
"""Run CLI command and return stdout as string."""
result = subprocess.run(
[self._executable, *args],
capture_output=True,
text=True,
env=self._env,
timeout=30,
)
if result.returncode != 0:
raise RuntimeError(
f"Command failed (exit {result.returncode}): {result.stderr}"
)
return result.stdout
def run_json(self, args: list[str]) -> Any:
"""Run CLI command expecting JSON output."""
output = self.run(args)
return json.loads(output)
```
### Import Mode Pattern (only with `--import` flag)
When the target is an installable Python package:
```python
"""Direct import client for {service} Python library."""
# Install the target: pip install {service-package}
from {service_package} import SomeClient, Config
class ImportClient:
def __init__(self, api_key: str):
self._client = SomeClient(Config(api_key=api_key))
def get_thing(self, thing_id: str) -> dict:
result = self._client.things.get(thing_id)
# Convert SDK objects to plain dicts for JSON serialization
return result.model_dump() if hasattr(result, "model_dump") else vars(result)
```
---
## Key Differences from TypeScript
| Aspect | TypeScript | Python FastMCP |
|--------|-----------|----------------|
| Tool registration | `server.tool(name, desc, zodSchema, annotations, handler)` | `@mcp.tool(description=...)` decorator on function |
| Parameter validation | Zod schema (compile-time) | Pydantic v2 model (runtime) |
| Error handling | Return `{ content: [...], isError: true }` | `raise ValueError("message")` |
| Startup | `await server.connect(transport)` | `mcp.run(transport="stdio")` |
| Logging | Use `console.error()` (not `console.log()`) | Use `print(..., file=sys.stderr)` |
| Auth check | `if (!API_KEY) throw new Error(...)` | `if not _API_KEY: raise ValueError(...)` |
---
## Error Handling Pattern
In Python FastMCP, raise exceptions with clear messages. FastMCP converts them to MCP error responses automatically.
```python
@mcp.tool(description="...")
def my_tool(params: MyParams) -> str:
try:
result = _client.do_something(params.thing_id)
return json.dumps(result, indent=2)
except httpx.HTTPStatusError as e:
raise ValueError(f"API error {e.response.status_code}: {e.response.text}") from e
except Exception as e:
raise ValueError(f"Unexpected error: {e}") from e
```
**Preferred behavior**: Raise explicit exceptions instead of catching everything silently or returning empty strings. The model cannot act on silent failures.
---
## Environment Variable Pattern
```python
import os
def _require_env(name: str) -> str:
"""Get a required environment variable or raise at startup."""
value = os.environ.get(name)
if not value:
raise ValueError(
f"{name} environment variable is required. "
f"Set it before starting the server: export {name}=your_value"
)
return value
# At module level (checked at startup, not at first tool call)
API_KEY = _require_env("SERVICE_API_KEY")
```
---
## `README.md` Template
```markdown
# {Service Name} MCP Server
MCP server for {Service Name}. Provides Claude Code with tools to {one-line purpose}.
## Tools
| Tool | Description |
|------|-------------|
| `service_get_thing` | Get a single thing by ID |
| `service_list_things` | List things with optional filters |
## Setup
### Prerequisites
- Python >= 3.11
- {Service Name} API key
### Install
\`\`\`bash
pip install -e .
\`\`\`
### Environment Variables
| Variable | Required | Description |
|----------|----------|-------------|
| `SERVICE_API_KEY` | Yes | {Service Name} API key from {where to get it} |
### Register with Claude Code
\`\`\`bash
python3 skills/process/workflow/references/mcp-pipeline-builder/scripts/register_mcp.py \
--name {service}-mcp-server \
--command python3 \
--args $(pwd)/src/main.py \
--env SERVICE_API_KEY=your_key_here
\`\`\`
Or add manually to `~/.claude.json`:
\`\`\`json
{
"mcpServers": {
"{service}-mcp-server": {
"command": "python3",
"args": ["/absolute/path/to/src/main.py"],
"env": {
"SERVICE_API_KEY": "your_key_here"
}
}
}
}
\`\`\`
```
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |