hono comes in transitively through @modelcontextprotocol/node, whose
range ^4.11.4 already allows the patched line, so only the lockfile
entry changes. 4.13.16 fixes GHSA-46wx-v44x-9hh2, GHSA-4m76-xfj5-7975,
GHSA-3gg2-335m-rxc6, GHSA-h7mg-cq63-vhpq and GHSA-4h7h-rv7c-7m2v.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Both helpers said "CI only" in their help text, but nothing enforced it. Run
from an elevated shell, or on a self-hosted runner, one would execute the
checked-out code as SYSTEM and the other would create a real local account
on a machine that outlives the run. They now stop unless GITHUB_ACTIONS is
true and RUNNER_ENVIRONMENT is github-hosted. The check stops an accident,
not an attacker.
The SYSTEM pass also removes its work folder (the runner script SYSTEM
executed and its output) in a finally block, including when the run fails.
A failed cleanup does not replace the outcome.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Below the sm breakpoint the Unmapped, Mapped and Ignored tables drop their
header and fixed column widths, which left the software name a few pixels
wide. Each row now wraps: the full name on its own line, publisher or vendor
below it, then the badges and the actions. The Map and Edit forms stack their
fields, and the tab bar spans the width so it fits with both count badges.
Every new class carries max-sm:, so the desktop table is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The reason text for no_dictionary pointed to the nightly sync only. The dictionary can be loaded on
demand, so the text names Vulnerability Data > Sync NVD Dictionary. The button's tooltip says "a few
seconds" instead of a number.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
A seventh button wrapped alone onto a second line. The dictionary sync sits next to the cache
refresh, both being NVD data, and is a filled button like the other syncs.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
- Data Sync Status: a line for the NVD product dictionary (when it was replaced, how many
products it holds, or that it is not loaded yet), and the cache line says that it counts the
products CVE data is held for.
- Button "Sync NVD Dictionary" next to the other syncs.
- The correlation switch sits on the line of its heading instead of the middle of the text block.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
- POST vulnerability/sync-dictionary (Global Admin) downloads the dictionary now. Until now only
the nightly run loaded it, so a new installation had none until 03:00 UTC and Auto-Map answered
that it was not loaded yet. 409 while a download runs on the instance.
- GET vulnerability/sync-status carries cpeDictionaryProducts and cpeDictionaryLastRefreshUtc,
both absent while no dictionary was ever stored. They are a different store than the CVE
cache that nvdCacheEntries counts.
- One dictionary download at a time per instance: the nightly run waits for one in flight and
then finds the stored copy current instead of pulling the feed a second time.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
The backend has read rules/dist/cpe-mappings.json since 1.5.1334; nothing reads
cpe-mapping-seed.json any more.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Its installer keeps the editor current on its own, like a browser. Findings stay listed in the
report; the headline numbers and ANALYZE-ID-003 leave them out. A Global Admin can change the
flag per product.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
- Session report: the badge of a row is the worst CVSS band over all CVEs of the entry again,
from the entry's counts on a newer report and from its list on an older one. The previous
commit had switched it to the report's risk level, which a known exploited CVE raises to
critical: rows of existing reports changed from HIGH to CRITICAL and contradicted the
severity cards above them.
- Admin > Software > Mapped: the actions column is wide enough for both buttons; Delete was
cut off at the table's edge.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
- get_vulnerability_summary: new parameter includeSelfUpdating (default false, like the portal).
The description names the default, the fields that say how much is left out, and that the
index holds per product and device every known exploited CVE plus the five most urgent others.
- search_sessions_by_cve: the description names the same limit of the index.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Session report
- The headline numbers come from the report's own counts, so a shortened CVE list no longer
changes them; each entry says how many CVEs it has in all next to the ones it lists.
- Self-updating software forms a group of its own below the findings and is left out of the
headline; the number of its CVEs stands next to it.
- Programs that could not be judged are listed under "Not evaluated" with the reason.
- A report written before these counts existed is read from its lists as before.
Software > Vulnerabilities and the dashboard card
- Self-updating software is left out by default. A line says how many products on how many
devices that is and counts them in on request; their rows carry a mark.
- The card says "None to update by hand" when only self-updating software was found.
Admin > Software
- Mapping and ignoring work on vendor plus canonical name.
- Automatic mappings show their source and can be deleted; a product can be marked as
self-updating.
- A Global Reader gets the lists without the write controls.
- The reseed result shows the inventory rows it changed and why a mapping list was refused.
Privacy and terms pages
- The NVD and EPSS lookups carry only public identifiers (a product identifier, a CVE number);
the NVD product dictionary is named as a downloaded source.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Mapping
- The backend keeps a reduced copy of the NVD product dictionary, replaced nightly when NVD
publishes a new one and read as a stream. The NVD keyword search is removed: no program name
is sent to NVD any more.
- A program gets an automatic mapping only when its name is exactly one product's name, the
product's cache row holds CVEs, none of them lacks a version, and the program does not occur
only as a Store package. Deleting an automatic mapping puts the name on the ignore list.
- rules/dist/cpe-mappings.json is the maintained list (410 mappings, editions, default set of
self-updating products). An import checks the list, writes, then removes what the list no
longer holds; a refused list leaves the mappings in service and is recorded as an ops event.
cpe-mapping-seed.json stays until this backend is deployed.
- The CPE on every inventory row follows the mappings: after each change and once a night.
- The admin routes work on vendor plus canonical name. POST vulnerability/cpe-product-flags
lets a Global Admin mark a product as self-updating; the value is stored only while it
differs from the list's default.
Report and rule
- Findings in self-updating software are listed, but the actionable_* counters leave them out.
ANALYZE-ID-003 (1.3.0) reads those counters, so a browser no longer triggers it.
- A program without a version number and a Store package under an automatic mapping are
listed as not evaluated instead of being compared.
- The fleet summary leaves self-updating software out unless includeSelfUpdating is set, and
says how much that is.
NVD
- A 403 is a refusal like a 429; a request that got no answer is told apart from a rejected one.
- Every loop that asks NVD has a count bound, an unanswered bound and a time bound. The cache
walk waits an open cooldown out once before its first request.
- The scan filter and the dictionary share one list of the platforms that rule a Windows
desktop out; the phone, embedded and ARM-only Windows lines are in it.
Monitoring
- New unsampled event SoftwareInventoryCpeReconciled; VulnerabilityDataSyncCompleted carries the
dictionary, the automatic mappings and the parts that fell short.
- Ops events VulnerabilityDataSyncDegraded (also from stored state: walk not finished for two
days, dictionary older than a week), CpeMappingUnavailable (mappings unreadable or a list
refused; once per UTC day, cause and severity) and CpeAutoMappingsCreated.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
get_software_inventory with scope "unmatched" advanced its cursor by the
page size asked for. The backend serves at most 200 rows per page, so a
larger page size skipped the difference without a hint. The cursor now
advances by the rows that arrived and ends on an empty page.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Tests cover the paged NVD fetch and its policy, the cache fill path,
the version filter (own product, editions, version schemes, platform
words), report entries and their trimming, the WhiteGlove merge, the
scan end to end over a mocked repository, the queue handler's
postponements, the re-scan, the per-session index replace, the refresh
walk, the KEV catalog and the write modes.
Fixes:
- A run that is out of time still reads the cache and only skips the
fetch. Before, every remaining product was reported as waiting for
data even when its data was cached.
- An entry of a report stored in the earlier format gets its counts and
the most-urgent-first list when it is merged. Before, trimming such an
entry left nothing that said how many CVEs it stood for.
The NVD client gets a test seam for its request pacing.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
CVE data
- The NVD fetch asked for one page of 100, which is the oldest 100, and
the age filter then dropped them. It now reads a full page and, for a
product with more, the last page as well, behind its own HTTP policy
and a time limit that covers the body. A timeout fails that request,
not the scan.
- A cache row keeps only the version ranges of its own product; the
filter compares the product again when reading, so rows written
earlier stop matching through another product's range.
- Known-exploited CVEs are kept ahead of the size cap.
- Cache rows and reports are written in replace mode. A merge left the
other storage form of an earlier write in place and the reader
returned it.
- Not judged, and reported as such: Store app stubs with a placeholder
version, ranges in another version scheme, and software editions a
mapping excludes. The platform check on descriptions matches whole
words.
- The KEV lookup reads the loaded catalog and never fetches; a failed
refresh is reported instead of swallowed.
Scan and result
- CVE data is read once per product. A product without data is fetched
while the run has time, otherwise noted for the nightly refresh; the
session is put back once and then reports it as not evaluated instead
of clean.
- A software entry lists every known-exploited CVE and the most urgent
others and carries the full counts; the report is trimmed before it
outgrows its table row. Sums and the WhiteGlove merge use the counts.
- New counters leave out software marked as self-updating.
- The queue handler and the re-scan store through one writer: the
report is replaced or removed and the session's CveIndex rows follow
it, keeping the detection date of rows that stay. The index takes the
five most urgent CVEs per software plus every known-exploited one,
and the fleet summary filters by date in the query.
- The nightly refresh uses the same fill path as the scan, starts with
products waiting for data and rows of the older shape, and removes
rows no mapping names.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Name patterns matched as substrings and the first hit won, so a short
pattern filed unrelated programs under a foreign product. The matching
now works on words:
- A pattern matches whole words in order; the pattern with the most
words wins. A one-word pattern needs the vendor to agree or the word
to be the whole name, and a name that follows "for" is an add-on.
- A key match compares vendor and canonical name (no version,
architecture or language), which becomes the key for manual mappings.
- The fuzzy NVD name search no longer runs during a scan. A program
without a mapping stays unmapped.
- Mappings load as one immutable snapshot: concurrent readers share one
load, a failed reload keeps the last good copy, and an empty result is
never served. A scan that cannot read them is put back on the queue
instead of filing every program as unknown.
- Every program is mapped before the first CVE lookup, so a run that
runs out of time has still filed each one. The inventory row takes the
computed CPE on every count, set or cleared.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
- Dashboard: a sixth stat card shows the known exploited CVEs (CISA KEV) of the last
30 days for the one tenant in view and opens Software > Vulnerabilities; it reads Off
while the Software Inventory & Vulnerability Analyzer is off. All stat cards share one
layout (icon in the title row, tighter padding); six cards sit in one row from a row
width of 1126 px, else 3 + 3 or 2 + 2 + 2, as container queries in globals.css.
- Software hub: a red known-exploited count on the Vulnerabilities tab. With the analyzer
off, the Vulnerabilities and Inventory tabs say so instead of an empty list; users who
can save the setting get a link that opens it with the switch highlighted.
- A session's Vulnerability Report and an app's detail page link to the tab; the hub takes
?tenantId= for a platform admin, like the app detail page.
- Telemetry: vuln_card_shown, vuln_link_clicked, vuln_enable_clicked,
software_tab_selected, vuln_exposure_viewed, vuln_exposure_action and
session_vuln_report_action. No CVE ids, session ids or product names.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The stored Admin Mode flag is shared by every account in a browser. It now
takes effect, and its toggle shows, only for a tenant Admin or a Global
Admin (canUseAdminMode); the stored value stays for the account that set it.
Dashboard rows offer delete and the Select checkbox only where the backend
admits the caller for the row's tenant (canAdministerTenant), the
blocked-device read runs only in the platform view, and a failed
Succeed/Fail shows an error notification.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
mark-failed, mark-succeeded and actions on sessions/{sessionId} take the
tenant from ?tenantId= (TenantScoping.QueryParam) and resolve it like the
other session write routes (ResolveSessionScopeAsync with
requireGlobalAdmin). Before, they used the caller's token tenant, so a
Global Admin on another tenant's session got 404. A test pins that every
sessions/{sessionId} route is query-scoped; the report keeps the tenant
from its body.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On the session page "/" focuses the timeline search from anywhere: the
controls row scrolls up below the navbar, so the matches fill the screen,
and a previous query is selected for overtyping. It is the web's usual
key for a page's own search and the same on Windows and Mac; Shift stays
allowed, since layouts like German type "/" as Shift+7. A "/" typed into
any field still belongs to that field. An empty box shows a "/" hint.
Esc in the box clears the query and keeps the cursor there for the next
term; Esc on an empty box leaves it.
The "is the user typing in a field" check moves to lib/keyboardTarget.ts
and is shared with the dashboard's Esc handler for select mode.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The ID stays retired; a test pins that it does not come back and that no rule
matches on the AutoLogon user name.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
The probe answered "gone" whenever it had no row, also when the query
itself failed. A console that still ran was then counted as script
activity instead of being reported. A start is now counted only on
certainty: the process list no longer has the id, it has another image
under it, or a fact that was read contradicts the start trace. A failed
or empty query, or a fact that could not be read, is reported at low
confidence, as before the counting was introduced.
The immediate upload is back for every oobe_console_spawned. The
confidence follows the command line, which whoever opens the console
chooses, so it labels the warning and never delays it. With short-lived
starts counted there is no burst left that the limit would have spared.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
The watcher learns of a process start about a second late, because WMI
delivers the start trace in batches. A scripted cmd /c has exited by
then, and its unreadable command line was reported as a low-confidence
console. A script that starts many of them produced one Warning with an
immediate upload each.
A start whose process is gone, or whose id now belongs to another process
(other image name, session 0, other parent), is now counted. The host
reports the count as one Info console_script_activity per 60 s window,
with the owner SIDs and the parent processes by image name, at most 20
windows per agent run and the rest when it stops. The type does not
count as activity, neither for the agent's idle and stall clocks nor for
the backend's stall heal.
oobe_console_spawned stays for a console that is still running when
probed. It carries parentProcessName and is uploaded immediately only at
high confidence.
Fixed in the same code: a reused process id could be reported with the
command line of another program; the owner of a console found at agent
start was never read, because the query lacked the key property; a later
console that was given an id seen before was dropped; the WMI
subscription was cancelled on the stop thread.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
WMI delivers the process start trace with a delay. By then the id can
belong to another process. The watcher took whatever ran under the id as
an Office worker, and a foreign process held the install window open
until it exited.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
ImeProcessIdentity carried a private wrapper around
QueryFullProcessImageName. It moves unchanged to
Monitoring/Interop/ProcessImagePath, so a second caller can read a
process's image name without a copy of the P/Invoke.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
The summary of the 410 response stood above ClampStartedAt and promised
that the agent treats registration and telemetry alike. It now sits on
WriteSessionLockedAsync and says what the agent does.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
register-session answers 410 while a deleted session's tombstone is fresh.
The agent treated that like a network failure: six attempts, exit 7,
everything kept for the next boot. A WhiteGlove resume never left that
loop, because the 48-hour break is skipped for it.
A 410 that echoes the request's correlation id is now its own signal. The
start sends one report (RegistrationRejectedSessionGone) and ends
monitoring with a forced cleanup, as the telemetry path already does on
410. A WhiteGlove Part 2 start rotates its session id and registers again
instead, so the user phase is observed; agent_started names the reason in
sessionRotationReason. A 410 without the echo stays a retried failure.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>