About BaseMcpServer
BaseMcpServer is a repository of Python MCP servers and the mcp-manager CLI for installing, configuring, and building them. It gives each server an isolated environment and can sync server entries to supported MCP clients.
Key features
- 01Includes multiple Python MCP servers
- 02Installs and manages servers with mcp-manager
- 03Creates isolated uv environments
- 04Syncs server configuration to supported clients
- 05Provides a server template and build guide
Use cases
- 01Install MCP servers from the repository
- 02Connect installed servers to MCP clients
- 03Create a new MCP server from the template
MCP support
The repository includes multiple MCP servers and mcp-manager for installing, managing, and syncing server configuration. It can configure connections for supported MCP clients.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEBaseMcpServer
A monorepo of Model Context Protocol servers plus the mcp-manager CLI that installs and configures them.
What's in this repo
BaseMcpServer/
├── utils/mcp_manager/ # CLI for installing + managing MCP servers (canonical install path)
├── servers/
│ ├── jira-helper/ # Jira + Confluence integration server (multi-instance)
│ ├── worldcontext/ # Current-context tools (date/time, market, news, dev-tool versions)
│ ├── mcpservercreator/ # Generator that scaffolds new MCP servers from a code snippet
│ ├── loadbearing-youtube/ # Extract a YouTube transcript + expose its load-bearing components
│ └── template/ # Starter scaffold to fork when building a new server
├── docs/
│ ├── adr/ # Architecture Decision Records
│ ├── developer/ # Implementation guides (build a new MCP server, client support, etc.)
│ └── user/ # End-user docs (Cline compatibility, etc.)
└── QUICKSTART.md # End-to-end jira-helper onboarding
Every server in this repo is a Python package installable via mcp-manager install, gets its own isolated uv-managed environment under ~/.config/mcp-manager/servers/<name>/.venv, and can be wired into supported MCP clients with a single mcp-manager sync call. The repository currently supports and tests macOS and Linux; Windows is explicitly out of scope.
Quick install
# 1. One-time: install uv and mcp-manager
curl -LsSf https://astral.sh/uv/install.sh | sh
uv tool install "git+https://github.com/dawsonlp/BaseMcpServer.git#subdirectory=utils/mcp_manager"
uv tool update-shell
# 2. Install whichever servers you want
mcp-manager install jira-helper --source ./servers/jira-helper
mcp-manager install worldcontext --source ./servers/worldcontext
mcp-manager install mcpservercreator --source ./servers/mcpservercreator
mcp-manager install loadbearing-youtube --source ./servers/loadbearing-youtube
# 3. Edit the per-server config files (where credentials go) — see each server's README
$EDITOR ~/.config/mcp-manager/servers/jira-helper/config.yaml
# 4. Wire the servers into installed supported clients
mcp-manager sync
For an end-to-end walkthrough with jira-helper (including Atlassian API token setup), see QUICKSTART.md.
Building a new MCP server
Fork servers/template/ and follow the README inside it. Every server uses MCP Python SDK v2 directly: a create_server() factory constructs MCPServer, then registers plain tool functions through add_tool(). Decorator-based registration and MCP Commons are not used.
Detailed reference: docs/developer/BUILD_A_NEW_MCP.md.
For dynamically generating a server from a code snippet, see the mcpservercreator server itself, once installed.
Connecting clients
mcp-manager sync # sync every detected supported client
# or selectively:
mcp-manager sync --platform cline
mcp-manager sync --platform codex
mcp-manager supports Cline, Claude Desktop, Claude Code, VS Code native MCP, Codex, and Antigravity. It writes or delegates stdio entries that point at the executable in each server's managed environment. Restart the affected client for new entries to take effect.
Example entry (written automatically; shown for reference):
{
"mcpServers": {
"jira-helper": {
"command": "/Users/<you>/.config/mcp-manager/servers/jira-helper/.venv/bin/jira-helper",
"args": ["stdio"],
"disabled": false,
"autoApprove": []
}
}
}
Hand-added entries in your editor's settings file are preserved across sync — only entries that match a server name in the mcp-manager registry are overwritten.
Architecture decisions
The major directional choices are tracked under docs/adr/:
- ADR-003 — current: all server installs go through
uv - ADR-002 — superseded; documents the earlier pipx-default attempt
License
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.