nicolargo/glances · error · RuntimeError
The 'mcp' package is required for MCP support. Install it wi
Error message
The 'mcp' package is required for MCP support. Install it with: pip install 'glances[mcp]'
What it means
GlancesMcpServer.__init__ raises RuntimeError when the 'mcp' python package could not be imported (MCP_AVAILABLE is False). MCP support is an optional extra, so the constructor hard-fails rather than serving a broken server.
Source
Thrown at glances/outputs/glances_mcp.py:73
MCP_DEFAULT_PATH = "/mcp"
def __init__(self, stats, args, config):
"""Initialize the MCP server.
Args:
stats: GlancesStats instance. May be None at construction time if the
stats manager has not been created yet; call set_stats() before
the server receives its first request.
args: Parsed command-line arguments (argparse.Namespace).
config: Glances configuration object (GlancesConfig).
"""
self._stats = stats
self._args = args
self._config = config
if not MCP_AVAILABLE:
raise RuntimeError(
"The 'mcp' package is required for MCP support. Install it with: pip install 'glances[mcp]'"
)
self._load_config(config)
self._mcp = FastMCP(
"Glances",
instructions=(
"Glances is a cross-platform system monitoring tool. "
"Use resources to access real-time system metrics "
"(CPU, memory, disk, network, processes, containers, sensors, …). "
"Use prompts to generate structured analyses of the current system state."
),
transport_security=self._build_transport_security(),
)
self._setup_resources()
self._setup_prompts()View on GitHub (pinned to a240d8dfb3)
Solutions
- Install the extra: pip install 'glances[mcp]'.
- Verify with python -c 'import mcp' — resolve any ImportError it shows.
- For Docker, use an image/requirements that includes the mcp extra.
Example fix
# before pip install glances python -m glances -s --mint # RuntimeError: The 'mcp' package is required... # after pip install 'glances[mcp]' python -m glances -s --mint
Defensive patterns
Strategy: validation
Validate before calling
try:
import mcp # noqa
mcp_ok = True
except ImportError:
mcp_ok = False
if not mcp_ok:
raise SystemExit("pip install 'glances[mcp]' before enabling MCP") Try / catch
try:
from glances.outputs.glances_mcp import GlancesMcpServer
server = GlancesMcpServer(stats, args, config)
except (ImportError, RuntimeError):
# MCP unavailable: skip MCP serving Prevention
- Pin glances[mcp] in requirements for MCP deployments.
- Smoke-test 'import mcp' in container entrypoints.
- Keep the extra installed across image rebuilds.
When it happens
Trigger: Constructing GlancesMcpServer (directly or via the --mint MCP/REST-MCP output path) in an environment where 'import mcp' failed — typically because glances was installed without the [mcp] extra.
Common situations: Running the MCP server mode in a pip install glances environment without extras; slim Docker images; CI environments testing MCP tooling.
Related errors
- JWT authentication is not available
- GlancesMcpServer: stats manager is not yet initialized. Call
- Plugin '{plugin}' not found. Available plugins: {available}
- Plugin '{plugin}' not found
- JWT authentication is not available. Install python-jose or
AI-assisted analysis of nicolargo/glances@a240d8dfb3 (2026-08-27).
Data as JSON: /api/errors/8d95d9c43ddd0fd6.
Report an issue: GitHub.