Mcp Server Builder — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Mcp Server Builder (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Model Context Protocol (MCP) server for searching MCP protocol and FastMCP documentation.
This MCP server helps you build correct MCP servers by providing searchable access to the official MCP specification and FastMCP framework documentation, always reflecting the current state of the protocol.
mcp, json, rpc, stdio) kept intactllms.txt sources on startupThe server indexes documentation from these curated llms.txt sources:
| Source | Description |
|---|---|
| modelcontextprotocol.io/llms.txt | Official MCP protocol specification |
| gofastmcp.com/llms.txt | FastMCP Python framework documentation |
uv from Astraluv python install 3.13Configure in your MCP client:
{
"mcpServers": {
"mcp-server-builder": {
"command": "uvx",
"args": ["mcp-server-builder@latest"],
"disabled": false,
"autoApprove": []
}
}
}Config file locations:
~/Library/Application Support/Claude/claude_desktop_config.json%APPDATA%\Claude\claude_desktop_config.json~/.codeium/windsurf/mcp_config.json.kiro/settings/mcp.json in your project# Using uv
uv add mcp-server-builder
# Using pip
pip install mcp-server-builderExample prompts to try:
search_mcp_docsSearch MCP protocol AND FastMCP framework documentation with ranked results and snippets.
search_mcp_docs(query: str, k: int = 5, source: str | None = None) -> list[dict]| Parameter | Type | Default | Description | |
|---|---|---|---|---|
query | str | required | Search query (e.g., "tool input schema", "stdio transport") | |
k | int | 5 | Maximum number of results to return | |
source | str \ | None | None | Optional filter: "mcp" for protocol docs, "fastmcp" for framework docs |
Returns: List of results with url, title, score, snippet, and source.
Examples:
# Search both sources
search_mcp_docs("how to define tools", k=3)
# Search only MCP protocol docs
search_mcp_docs("lifecycle", source="mcp")
# Search only FastMCP framework docs
search_mcp_docs("authentication", source="fastmcp")fetch_mcp_docRetrieve full documentation page content by URL from MCP protocol or FastMCP framework docs.
fetch_mcp_doc(uri: str) -> dict| Parameter | Type | Default | Description |
|---|---|---|---|
uri | str | required | Document URL (http/https from supported domains) |
Supported domains: modelcontextprotocol.io, gofastmcp.com
Returns: Dictionary with url, title, content, source (or error on failure).
Example:
# Fetch MCP protocol doc
fetch_mcp_doc("https://modelcontextprotocol.io/docs/concepts/tools")
# Fetch FastMCP framework doc
fetch_mcp_doc("https://gofastmcp.com/tutorials/tools")Step 1: Search for relevant documentation
search_mcp_docs("FastMCP tool decorator", k=5)Step 2: Fetch full content of the most relevant result
fetch_mcp_doc("https://gofastmcp.com/tutorials/tools")llms.txt URLs (MCP + FastMCP)
↓ startup
BM25 index with stemmed tokens + n-grams (titles only, fast)
↓ search request
Stem query → match unigrams/bigrams/trigrams → BM25 rank
↓ top-k
Lazy content hydration → snippet extraction
↓ response
{ url, title, score, snippet }# Clone and install
git clone https://github.com/praveenc/mcp-server-builder.git
cd mcp-server-builder
uv sync --dev
source .venv/bin/activate
# Run tests
uv run pytest
# Run with MCP Inspector
npx @anthropic-ai/mcp-inspector uv run mcp-server-builder
# Linting and type checking
uv run ruff check src tests
uv run pyrightMIT - see LICENSE for details.
Contributions welcome! Please open an issue or submit a pull request.
For issues and questions, use the GitHub issue tracker
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.