MCP server for NetLogger: nets on the air now, live check-in lists and who's up, and past nets and their check-ins, through any MCP-compatible AI assistant.
Data from NetLogger's public XML Data Service (API 1.3). Part of the qso-graph project. No API key needed. Your callsign is asked for once (see below).
uvx netlogger-mcp # run it; nothing to install
pip install netlogger-mcp # or install it into your own environment| Tool | Description | Key Parameters |
|---|---|---|
netlogger_active_nets |
Nets on the air now: frequency, band, mode, net control, logger, monitoring count | name_like |
netlogger_checkins |
A live net's check-in list, plus the pointer (the station being worked now) | server_name, net_name |
netlogger_past_nets |
Closed nets over the last N days, with the net IDs past check-ins need | interval_days, name_like |
netlogger_past_checkins |
A closed net's check-in list | server_name, net_name, net_id |
netlogger_set_callsign |
Save your callsign (asked once, on first use) | callsign |
get_version_info |
Service version + upstream spec version (fleet identity attestation) | — |
NetLogger is the logging program many amateur-radio nets use: net control or a logger keeps the check-in list, and anyone can follow the net live. Its server publishes active nets, check-in lists and past nets through a public API.
Every request tells NetLogger which station is asking, so it can tell one user from another:
netlogger-mcp/0.1.0 (KI7MT; +https://github.com/qso-graph/netlogger-mcp)
On first use the assistant asks for your callsign and saves it. You're asked once. To set it yourself, use NETLOGGER_MCP_CALLSIGN=KI7MT.
It's saved in settings.json, in ~/.config/netlogger-mcp/ (Linux), ~/Library/Application Support/netlogger-mcp/ (macOS) or %APPDATA%\netlogger-mcp\ (Windows).
NetLogger is a donation-funded service on one server, and it limits how often each call may be made. We keep to those limits:
| Measure | Detail |
|---|---|
| NetLogger's call limits | GetActiveNets 1/min, GetCheckins 3/min, GetPastNets 1/min, GetPastNetCheckins 10/min. Checked before a request is sent, never after. |
| One budget per user | Every copy you run (Claude Desktop, Claude Code, a script) shares one budget, through a locked file beside settings.json. |
| Shared answers | Every copy also shares the answers it fetched, so a copy that has hit the limit gets another copy's recent answer, with its age, instead of an error. |
| Response caching | Active nets 60 s, check-ins 20 s, past nets 60 s, past check-ins 1 hour. Filtering by name costs no extra calls. |
| How old an answer is | Every answer has as_of_utc (when it was given) and age_seconds (how old the data is), so it was fetched at as_of_utc minus age_seconds. |
| Stale answers over errors | When a limit is reached, the last answer comes back with its age rather than a new request. |
| 429 back-off | A "too many requests" on any call stops all calls for at least a minute, longer if NetLogger asks. |
| Past-net windows | More than 7 days of past nets needs a name filter, NetLogger's rule to protect its server. |
| Request timeout | 15-second timeout. |
| User-Agent header | Every request names this project and your callsign, so NetLogger's operators can see who is asking. |
NetLogger's band and frequency are whatever the logger typed ("40M", "75M", "DMR", "7033", "3918 KHz", "147.030 91.5", "TG 3141"). They're returned as typed, and beside them, only when the text can be read reliably: frequency_mhz, band_adif (the ADIF band, from ADIF 3.1.7's Band enumeration), tone_hz and talkgroup. A frequency outside the amateur bands (e.g. GMRS) gets no band_adif.
NetLogger's check-in data includes street addresses and ZIP codes, and past nets include the IP address of whoever opened them. None of these are ever returned.
netlogger-mcp works with any MCP-compatible client. Add the server config and restart. The tools appear automatically.
Add to claude_desktop_config.json (~/Library/Application Support/Claude/ on macOS, %APPDATA%\Claude\ on Windows):
{
"mcpServers": {
"netlogger": {
"command": "uvx",
"args": ["netlogger-mcp"]
}
}
}Add to .claude/settings.json:
{
"mcpServers": {
"netlogger": {
"command": "uvx",
"args": ["netlogger-mcp"]
}
}
}{
"mcpServers": {
"netlogger": {
"command": "uvx",
"args": ["netlogger-mcp"]
}
}
}Add to .cursor/mcp.json (project-level) or ~/.cursor/mcp.json (global):
{
"mcpServers": {
"netlogger": {
"command": "uvx",
"args": ["netlogger-mcp"]
}
}
}Add to .vscode/mcp.json in your workspace:
{
"servers": {
"netlogger": {
"command": "uvx",
"args": ["netlogger-mcp"]
}
}
}Add to ~/.gemini/settings.json (global) or .gemini/settings.json (project):
{
"mcpServers": {
"netlogger": {
"command": "uvx",
"args": ["netlogger-mcp"]
}
}
}Installed with pip instead? Use "command": "netlogger-mcp" in any config above.
"What nets are on the air right now?"
"Are any 80m nets running?"
"Who's checked into the county ARES net, and who's up now?"
"Which nets ran in the last three days?"
"Show me the check-ins from last night's net."
The same code works without an AI, with the same limits and cache. A library can't ask for your callsign, so you pass it in:
from netlogger_mcp.netlogger import NetLoggerSource
nl = NetLoggerSource(callsign="KI7MT")
for net in nl.active_nets()["nets"]:
live = nl.checkins(net["server"], net["name"])
print(net["name"], "up now:", live["pointer_callsign"])Apps can also name themselves, using ADIF's PROGRAMID and PROGRAMVERSION: NetLoggerSource(callsign="KI7MT", program_id="MyLogger", program_version="1.0").
NETLOGGER_MCP_MOCK=1 netlogger-mcpnetlogger-mcp --transport streamable-http --port 8014git clone https://github.com/qso-graph/netlogger-mcp.git
cd netlogger-mcp
uv sync --group dev
uv run pytestGPL-3.0-or-later. adif_band.json is the ADIF 3.1.7 Band enumeration, the work of the ADIF Developers Group (adif.org.uk), included unchanged.