Repository navigation
Releases
v0.3.1
v0.3.0
Added
- A
503carryingRetry-Afteris now honoured as a limit. A429always
was; a503was treated as an outage whatever it carried, so NetLogger
shedding load deliberately and saying when to come back was answered sooner
than it asked. A bare503is still an outage, because the site may come back
in seconds and an outage lets the next request through.
Changed
-
Python 3.11 is the minimum. 3.10 reached end of life in October 2026
(PEP 619), so it gets no further security
releases on any platform. On Windows it was worse than unsupported:uv's 3.10
is built against OpenSSL 1.1.1w, end of life since September 2023, so HTTPS
from this server ran over an unmaintained TLS library.certifidoes not help
with that — it supplies the root store, not the implementation.3.10 support was already nominal. This package has linted with
ruff target-version = "py311"and type-checked withmypy python_version = "3.11"for some time, so nothing ever verified the 3.10 claim.uv.lockis regenerated and the diff is large, but nothing is upgraded —
it is the 3.10 resolution branch disappearing. Packages that carried two pins,
one per Python branch, now carry the single 3.11+ pin they already resolved to.
An install on 3.11 or later is unchanged. -
The Windows entry in the test matrix moves from 3.10 to 3.11, so Windows is still
covered at the lowest supported version rather than losing coverage.
v0.2.2
Changed
--transport streamable-httpnow says where it bound, and the help text says
it too. It binds127.0.0.1, and FastMCP takes no host option, so it was
never reachable from another machine or from outside a container — and nothing
said so. The bind address is unchanged: loopback-only is deliberate for a
server that holds credentials and has no authentication of its own.
v0.2.1
- A release should appear on the Releases page
- Declare Semantic Versioning, like the other 20 do
- Move every Node 20 action to its first Node 24 major
- The locked-dependency audit blocked every release
- Four upgrades arrived as 57 pull requests
- Nothing was auditing the lockfile
- Close every open security advisory in the lockfile
- A release is a workflow, not a merge
- Bump certifi from 2026.7.22 to 2026.10.19
- Make the publish gate's http:// check portable
v0.2.0
The minor version moves because two things a caller may have relied on now behave differently.
Upgrading: a tool that fails now returns isError: true instead of a successful result whose
payload carried an error key. Anything reading result["error"] should read the error flag.
The one that will be noticed: needing a callsign used to be a successful result carrying
needs_callsign: true and a next_step. It is now a flagged error whose message is the whole of
that instruction. A caller that read needs_callsign should instead read the error flag; the text
still says exactly what to do and names netlogger_set_callsign.
MCP specification 2026-07-28 (#46)
- Tool errors are flagged
isError(#46). Five error returns insrc/arrived on the wire as
successful results, so a client could not tell "the tool answered" from "the tool refused". The
specification says ofCallToolResult.isErrorthat otherwise "the LLM would not be able to see
that an error occurred and self-correct" — we were relying on the model reading the word error
in prose where the protocol has a flag. They now raiseToolError, which FastMCP flags with the
message intact. - The catch-all branch deliberately does not pass
str(e)through. It exists for what was not
foreseen, and an unforeseen exception's text is not known to be safe to show; the three branches
above it carry the messages that were written to be read. - Every tool declares all four annotation hints (#46). Omitting them is not neutral: the
defaults advertise a tool asdestructiveHint: true,idempotentHint: falseand
openWorldHint: true, and the specification tells a client it may keep a human in the loop for
an operation like that. So six read-only-looking tools were published as potentially destructive. netlogger_set_callsignisreadOnlyHint: false, and that is the point. Issue #46 says
"every tool in this repo is read-only" and that is not so — this one writes the operator's
callsign to a settings file. It is the first tool in this fleet sweep that is not read-only, and
copying the block the other nine servers use would have published a plain untruth about it. It is
destructiveHint: false(it replaces one value the operator set themselves, and loses nothing
else) andidempotentHint: true(saving the same callsign twice leaves the same state), so a
client may safely retry it.openWorldHint: false: it writes locally and tells NetLogger nothing
until the next lookup.get_version_infoisopenWorldHint: false— it reports versions carried in the wheel and opens
no connection. The four NetLogger lookups aretrue.tests/test_mcp_conformance.pyasserts all of it, through a realfastmcp.Clientwhere the
claim is about the wire. Each check was falsified once — the check removed, the suite watched to
fail — including the two that matter most here: anetlogger_set_callsignthat claims to be
read-only, and a refusal that keeps the flag but loses the instruction.get_version_infodrifted
across this fleet precisely because it was a convention nobody asserted.- Two controls back the write hints with behaviour rather than trust: the tool really does write a
settings file, and a refused callsign leaves no file behind. as_of_utcwas on the error payloads as well as the successful ones. It is unchanged for a
successful answer, which gets it from the source layer, and a test says so.
TLS (#44, qso-graph-devel#73)
certifiis carried, and the fetch verifies against it (#44, qso-graph-devel#73). The one
urlopeninnetlogger.pypassed no context, so it trusted whichever root store the running
Python was built against. On macOS that is/etc/ssl/cert.pem, a deliberately reduced
OpenSSL-compatibility file: current, not stale, but it omits roots the OS itself trusts, so a
certificate chaining to one of them fails for Python whilecurlto the same URL succeeds.- Measured, not assumed:
www.netlogger.orgchains through Let's Encrypt'sYR1to ISRG
Root YR, and verifies against Apple's file and againstcertifialike today. So this was
latent rather than an outage. What it buys is that the root set is now a dependency this project
updates, rather than a file the operating system hands us — and the same onerequestsand
httpxwould have used had this fleet not been written against stdliburllib. certifiholds fewer roots than Apple's file (121 against 128), so the point is not a bigger
set. It is the right set, chosen by Mozilla's root program.- This server holds no credential — a callsign is public — so a trust failure costs an answer
rather than a secret. But it would cost it invisibly: the user is told only that NetLogger
couldn't be reached, with nothing pointing at trust. Verification is never disabled;
ssl.CERT_NONEwould make a call "work" by accepting any certificate from anyone, which is a
worse defect than this one and a silent one. certifiis declared in[project] dependencies, and a test says so. Measured: it is not
a transitive dependency offastmcp>=4.0,<5+defusedxml— that pair resolves 85 packages and
certifiis not among them. An undeclared import here would be an ImportError at startup for
anyone who ranpip install netlogger-mcp, so this is a hard failure rather than a latent one.- Four static checks in
tests/test_security.py(verification never disabled, the root set carried
not borrowed,certifideclared, every outbound call verifying) and one behavioural check in
tests/test_netlogger.pythat watches the context arrive aturlopenand asks what it will do.
The behavioural one exists because a grep cannot tell a verifying context from one built with
verification switched off — falsified exactly that way, and it caught it. test_every_outbound_call_verifiesscans the whole package line by line, not one file: there is
one fetch here today, and a check written against the file you happen to be editing is a check
that misses the second one. Falsified by adding a second fetch path with no context.
Lint (#45, qso-graph-devel#70)
E501is adopted (#45, qso-graph-devel#70). It reported 105 lines, and not one of them
was a defect — which is the finding, not an argument against enforcing it. A consistent width
is worth having; it just was never the thing holding the quality. The rules that do find defects
(F,I, mypy strict) went in first, which is why this waited.- One line is exempt, at the line rather than the file, and the config names what is exempted:
tests/test_netlogger.pyasserts NetLogger's own query string whole, because the test is that we
send exactly that. Split across two literals it stops being greppable against NetLogger's
documentation, which is the reason for asserting the URL instead of its parameters. There the
length is the correctness. It is the only# noqa: E501in the repo, and removing it makes
rufffail — so it is needed rather than decorative. USAGEnow opens with a line continuation so the synopsis sits under the width without being
wrapped. The printed--helptext is byte-for-byte what it was (same SHA-256): wrapping a usage
synopsis would be a change to what a user reads, not formatting.- Two things on the wire change, and only these two. Reflowing an
Args:line puts a newline
into the parameter description FastMCP publishes ininputSchema, so
netlogger_active_nets.name_likeandnetlogger_past_nets.interval_dayseach gain a\n. The
tool schemas were dumped through a real client before and after, and nothing else differs. Said
here because the rest of this is formatting, and a reader is entitled to assume formatting
changes nothing.
Docs
- The README's User-Agent example said
netlogger-mcp/0.1.9directly above the sentence "the
version is the installed one — it is read from the package, not written here". 0.1.10 set out to
stop freezing a number there and froze a newer one instead. The example now reads
netlogger-mcp/<version>, which is what the sentence claims and cannot go stale.
v0.1.10
- ruff and mypy run in CI (qso-graph-devel#66), as a job the
ci-all-greengate requires.
Settings followadif-mcp, the reference for every qso-graph Python repo, rather than a style of
this repo's own. They run once rather than per Python version: both read the source, and neither
answer changes with the interpreter. - The inter-process file lock tests
sys.platformrather thanos.name. Both are true on
Windows, but only the first narrows for a type checker, so the Windows branch was being read
against a POSIX stdlib with nomsvcrtin it. - The shared limiter and cache files name what their dicts and lists hold,
_runtypes its
*args, and a loop variable that shadowed the note read out of the same check-in record is
renamed. Four test lines carry one statement each. E501is deferred rather than adopted (qso-graph-devel#70): what it reports in these repos are
widths, not defects, and some lines are long because they name a publisher's field exactly.mcp.runis given the literal fastmcp asks for rather than astrthat happens to hold the
right word.fastmcpis bounded:>=4.0,<5(qso-graph-devel#60). It was>=3.0with no upper bound, and
these servers are run withuvx, which resolves fresh — so afastmcp5.0 would have reached
every user automatically, before anything here had been run against it. The floor rises to 4.0
because that is what is actually tested: every lock in the fleet held a 4.x, and nothing in CI
has ever exercised 3.x. A claim of 3.x support that no test backs is not support.fastmcpis locked at 4.1.0, the current release, so CI runs against what a new install gets.- The published contact is
maintainers@qso-graph.io(qso-graph-devel#69). Theauthorsfield
carried a personal address, and that field is what PyPI shows on the package page. Everything in
qso-graph is open source and open to contribution, so the contact is the project's. - The README's User-Agent example named 0.1.0 while the package was 0.1.9. The version in it is
read from the installed package, so the example says so rather than freezing a number that
drifts every release.
v0.1.9
v0.1.8
- PyPI: the Documentation link goes to this package's own page, https://qso-graph.io/servers/netlogger-mcp/ (qso-graph/.github#15).
- CI: the release flow (qso-graph/.github TEMPLATES.md). Work lands on
develop; a release is a
PR fromdevelopintomain, and merging it publishes to PyPI and the MCP Registry, verifies both
and tags the release. CI runs ondeveloptoo, and PRs intomainmust come fromdevelopor a
security/branch.
v0.1.7
scripts/anonymize_samples.pynow also finds callsigns that begin with a digit (4X4ABC,
2E0ABC,3D2AB,9A1AA) or are in lower case, in free-text fields, while leaving a D-STAR
reflector's35C, grid squares, frequencies andC4FMalone (#29). Test-tooling only: the
published package is unchanged, and today's samples regenerate identically.
v0.1.6
- Tested against real NetLogger responses (#15). The bundled samples (mock mode) are now one
real response per routine, captured within NetLogger's published rates and anonymized by
scripts/anonymize_samples.py: callsigns, names, places and remarks replaced (callsigns with
N0CALL-style ones), grids cut to 4 characters, street, ZIP and source IP set to test values. The
earlier synthetic samples move totests/fixtures/spec/, kept for what the real responses didn't
show (a<Warning>, a non-UTC time zone, an unknown element, club member IDs). - The parser held up: no behaviour change was needed. What the real responses showed, now
pinned bytests/test_real_responses.py:- the server sends
InactivityTimer(the spec also spells itInactivitytTimer; both are read); - past nets list their elements in a different order from the spec (order doesn't matter here);
- no
<Warning>in any header; MemberIDis present but empty on nets that aren't a club's;- role markers as sent:
(nc),(log),(m),(c/o); - free-text frequencies (
all,REF 35C,7.185.5,14. 290, even a callsign) are kept as sent
and not read as a frequency; - an empty net control, and a closed net's pointer one past its last row.
- the server sends