Troubleshooting
Server stays offline / red / error
- Run the same command your host uses:
bash
# activate your venv first
python -m wireshark_mcp doctor
- If doctor fails, install Wireshark/TShark or set
TSHARK_PATH. - Run
python -m wireshark_mcp setupand reload the host. - Confirm
commandis an absolute path to the venv Python that has this package installed.
This server exits on startup when TShark is missing. That is intentional: hosts treat a failed process as offline instead of advertising broken tools.
tshark not found
- Install Wireshark so
tsharkexists. - Or set in MCP config:
json
"env": {
"TSHARK_PATH": "C:\\Program Files\\Wireshark\\tshark.exe"
}
- On PATH-only installs, ensure the host process inherits PATH (GUI apps on macOS/Windows often do not). Prefer
TSHARK_PATH.
Live capture fails, file analysis works
- Windows: install Npcap and reboot if prompted.
- Use
list_interfacesand pass an exact interface name. - Capture only on networks you are authorized to inspect.
- Prefer short durations (3–10 seconds).
Tools error with event-loop / PyShark messages
Current tools drive TShark directly. Reload the MCP server after upgrading so the host is not running an old process.
Wrong Python / module not found
/absolute/path/to/.venv/Scripts/python.exe -m wireshark_mcp doctor # Windows
/absolute/path/to/.venv/bin/python -m wireshark_mcp doctor # Unix
Reinstall into that interpreter:
python -m pip install -e .
Permission denied capturing
Unix: add your user to the wireshark group (distro-specific) or run with appropriate capabilities. Prefer analyzing saved pcaps when possible.
GitHub Pages site for these docs
The public docs site is built by the Pages workflow. If deploy fails with Resource not accessible by integration, an admin must set Settings → Pages → Source = GitHub Actions once, then re-run the workflow. The Node 20 line in Actions logs is a runner warning, not the failure cause.