Repository navigation
Releases
v1.16.0 — Investigation brief
Investigation brief — 1.16.0 build 113
Candidates, Review Library and Dependency Explorer now answer four questions for
the selected object: why review, what is known, what remains uncertain, and what
to check next. One explicit suggestion opens the appropriate evidence, with the
same bounded facts available in dedicated CSV and JSON exports.
- Keep saved stages, evidence validity, usage and protection independent.
- Label missing/historical evidence and an unverified review store as unknown,
not zero dependencies, no decision, or proof of removal. - Open changed/unverifiable Library reviews on evidence while preserving tab
choices, notes and draft navigation guards. - Make Explorer’s brief and Changes view follow the inspected identity, not
an unrelated graph center. Explicit relationship suggestions reset display scope. - Clarify cached collection timestamps. Preserve existing full report formats,
dependency membership, scoring, coverage and removal-readiness rules.
No customer-content writes, automatic collection, new permissions or review
migration. Customer content remains read-only. No shared restart or new scan is
required solely for this feature. Existing review/policy journals and baselines
are preserved.
Compatibility and scope
Validation: 1,318 automated checks; zero pre-cert AppInspect errors/failures;
14 schemas; 6,892 real cached-object summary checks; 38 accepted native captures;
28 JSON and 28 CSV download comparisons. Existing and historical Library cases
were exercised where already present; the other two Libraries stayed empty.
Original review/policy journals, inventory manifests and appearance preferences
are unchanged. Interrupted automation captures are excluded.
AppInspect retains three existing reviewed advisories and one manifest skip;
the existing trusted-library cache is reused.
Minimum: Splunk Enterprise 10.2.5 on a standalone Linux x86-64 search head with
healthy KV Store. Light/Dark qualification targets Enterprise 10.2.5, 10.2.8 and
10.4.4. This is bounded static investigation, not runtime dependency closure,
complete consumer coverage, or removal approval. Cloud, clustered/Windows review
writes, every viewport/control and production latency remain outside this gate.
See the README, user guide, compatibility notes and troubleshooting for the
workflow and its limits. Private validation reports, lab details and credentials
are not release assets.
Archive
content_hygiene-1.16.0.tar.gz
SHA-256: 723559cc4ce93a58fd9612bdbb1e12b8652ff2ef50e5ee2f6d6f284cf99bbb6b
Portable USTAR, with macOS sidecars/metadata, tests, source maps, local state and
development material excluded. Attachments: the archive and SHA256SUMS only.
v1.15.0 — Actionable caller-context diagnostics
Actionable caller-context diagnostics
Dependency Explorer and Review Library now explain why captured search/dashboard
queries retain or withhold initial caller context. Investigators can inspect exact
query locations and digests, first-rejection reasons, independent syntax
observations, source provenance and concrete next verification steps.
- Filter center/related objects by withheld or unrecorded caller context.
- Search, filter and page through bounded query diagnostics using shared,
wrapping Splunk-native controls. - Export all retained diagnostic facts as JSON or CSV, independently of display
filters, including explicit older, empty, unavailable and limited states. - Keep customer content read-only and preserve existing review decisions,
notes, baselines, revision history, policies, confidence and scoring.
Upgrade and compatibility
Version 1.15.0, build 112. Minimum support remains Splunk Enterprise 10.2.5
on a standalone Linux x86-64 search head with healthy KV Store. The exact archive
was installed and its targeted Explorer/Library workflows checked in Light and
Dark modes on Enterprise 10.2.5, 10.2.8 and 10.4.4, including modern navigation.
No shared restart, review migration or new source permissions were required.
Run an explicit inventory scan to capture the new diagnostic facts. Existing
snapshots remain usable and show Not recorded rather than implying complete
coverage. Merely opening a page never launches a scan.
Verification
- 1,267 automated checks passed, with lint/style/type/build/package gates.
- Pre-cert AppInspect 4.3.1: zero errors, failures or future failures;
three existing reviewed warnings and one manifest skip. The trusted-library
cache was reused; this is not Splunk Cloud certification. - Strict-parser, analysis and review-evidence parity passed for 2,209 captured
shared caller sources on identical inputs. - 13 schemas, 2,209 fresh native caller summaries and 25 JSON / 25 CSV
downloads validate against cached facts and configuration provenance. - Original review/revision/policy journals are unchanged and appearance
preferences restored. No customer objects, accounts or validation reviews
were created or edited. - Functional tests compile their real lazy-route modules before timed
assertions, avoiding a reproduced Jest/Babel cold-start readiness failure.
Production lazy loading, mocks, assertions, timeouts and the tested archive
are unchanged. The initial CI failure remains in the workflow history.
Deliberate limits
This is an investigation/accuracy-explanation feature, not broader parser
coverage or proof of runtime execution. It does not expand macros or Boolean
grammar, create new dependency conclusions, strengthen confidence or establish
removal eligibility. Post-process/chain queries are not independent original-event
callers. Counts describe extracted query occurrences, not all consumers.
Diagnostics retain at most 64 rows / 64 KiB per object and 8 MiB of detailed
facts per scan, with explicit count/limit markers. Raw SPL, selector values and
macro definitions are absent from these facts. Source extraction, private/runtime
content and unsupported syntax remain uncertain. Qualification is targeted
desktop testing, not every control/viewport, mobile, production-latency, Cloud
or clustered/Windows coordinated-write certification. The approximately 1.53 MiB
bundle advisory remains.
See the user guide,
compatibility notes
and troubleshooting.
Download integrity
content_hygiene-1.15.0.tar.gz
SHA-256: 91d7c7d6082b61cd60c255e610346ae400e77010effac219feaab41359d799c8
The archive has one content_hygiene/ root, normalized portable tar headers and
no macOS sidecars, credentials, customer/lab data, local state or development
reports. Install this archive; SHA256SUMS is provided separately.
v1.14.0 — Context-aware alternative paths
Context-aware alternative-path investigations
Build 111 adds an explicit, read-only way to compare a chosen dependency witness
with alternative directed paths through the existing cached graph.
- Use Compare alternative paths in Dependency Explorer, In-use Content and
Removal Impact. Review Library adds Dependency paths with exact endpoint
selection and direction, without changing the saved decision. - Keep the original witness and inspect up to three alternatives independently.
Caller, app, query and branch evidence is never combined across paths. - Select readable native path buttons, inspect ordered hop evidence/provenance,
focus every exact path connector/object in Explorer, and return to the original
map, filters, pages and camera. - Export dedicated path-investigation CSV/JSON with all displayed paths,
diagnostics and limits. Existing full reports retain their original scope. - Load and integrity-check Review Library's cached graph, preventing a
summary-only page load from producing a false empty investigation.
Qualification and compatibility
The exact archive passes 1,230 automated checks, lint/style/types, build and
portable-package gates. Pre-cert AppInspect 4.3.1 reports zero errors, failures
or future failures; three existing reviewed advisories and one manifest skip
retain their documented dispositions.
Enterprise 10.2.5, 10.2.8 and 10.4.4 pass all seven pages in Light/Dark at
1280 × 720, with 84 accepted native captures and 32 JSON / 32 CSV download
comparisons. Original cached catalogs, manifests, decisions, baselines and
review/policy journals are preserved. Appearance preferences are restored.
No customer-content writes, source execution/sampling, new scans, reviews,
policies, accounts or shared restarts are required by this increment.
Search is bounded to six hops, 20,000 edge preparation/visits, 2,048 states,
64 completed candidates and three displayed alternatives. An original witness
can retain up to 32 hops. No result is proof of absence or removal eligibility;
display order is not a global best-path or execution ranking. Positive
shared-caller alternatives and positive Library paths remain automated contracts
where the available native catalogs lack examples. Wide paths may require
horizontal panning. Narrow/mobile panes, Cloud, clustered/Windows coordinated
writes and production-latency/runtime-closure certification remain outside scope.
See the investigation guide
and qualification, limits and recovery notes.
Install the attached content_hygiene-1.14.0.tar.gz; verify its SHA-256 against
the attached SHA256SUMS. Minimum supported runtime remains standalone Linux
Splunk Enterprise 10.2.5 with healthy KV Store. MIT licensed.
v1.13.0 — Navigable dependency maps
1.13.0 — Navigable dependency maps
Content Hygiene remains read-only against customer content. Existing dependency
evidence, review decisions and policy journals are preserved.
- Hover or keyboard-focus a relationship box to emphasize its exact connection
and endpoint cards. Full context keeps names, evidence and source locations readable. - Expand/collapse the whole map or one branch. High-degree columns use bounded
object pages and explicit continuation groups instead of a very tall stack. - Search captured connections and focus both exact endpoints together, including
off-page/grouped objects. Return to the previous map without changing its center
or filters. Full CSV/JSON reports do not shrink to the displayed pair or page. - Use adaptive zoom/detail and a full-window workspace with native Splunk
Light/Dark controls, keyboard dismissal and focus restoration.
The exact build 107 archive passes 1,194 automated checks and pre-cert AppInspect
with zero errors/failures. All seven pages pass desktop Light/Dark validation on
Enterprise 10.2.5, 10.2.8 and 10.4.4, with 120 accepted native captures and 12 JSON /
10 CSV comparisons. Original real caches, decisions, baselines and policy journals
retain their hashes. No rescan or shared restart is required for this UI update.
Maps remain bounded to three hops, 60 source cards and 20,000 expansion visits.
Wide maps can require horizontal panning. Desktop qualification is at 1280 × 720;
very narrow embedded/mobile panes are not qualified. See the
guide
and qualification/recovery limits.
Install content_hygiene-1.13.0.tar.gz and verify it with the attached SHA256SUMS.
The archive excludes private reports, development state, credentials and macOS metadata.
SHA-256: 19a8576dd24b3f03734c1eb3b91b85a3bf277fbcc22e9185be0fb70381420e81
Splunk Content Hygiene 1.12.0 (build 100)
Splunk Content Hygiene 1.12.0 — build 100
Context-aware witness paths make multi-hop dependency investigations easier to
interpret without weakening the conservative dependency inventory.
Changes
- Explorer graph details, required-content and removal impact distinguish shared
captured caller receipts, app-only evidence, mixed, unverified, bounded and
static-only paths. Independent queries, panels, roots and branches stay separate. - Uncertain-hop links open the exact cached object; the same compact context and
ordered witness evidence is carried into CSV/JSON exports. - Full-width impact witness details support keyboard selection and return, with
close and result-pagination controls remaining visible. - Unknown reusable invocation never becomes a false fixed-app mismatch.
- Dependency membership, confidence, impact scores, readiness, collection
coverage, existing decisions, baselines and policy journals remain unchanged.
Compatibility and validation
- Minimum: Splunk Enterprise 10.2.5 on a standalone Linux x86-64 search head,
healthy KV Store and bundled Python. Representative native Light/Dark checks
cover Enterprise 10.2.5, 10.2.8 and 10.4.4. - 1,153 automated checks plus lint, style, types, production build and portable
package verification. Pre-cert AppInspect 4.3.1: zero errors, failures or future
failures; three existing reviewed advisories and one manifest skip. - All seven pages in both themes, 72 accepted native captures, 104 prior-release
root comparisons, 28 JSON / 32 CSV actual-download comparisons, and validation
of all 11 schemas against 21,104 cached object/edge/scan records. - Original review/policy journals, cached manifests, captured baselines and native
appearance preferences preserved. No customer-content writes, new reviews,
accounts or policies, rescans, expression execution/sampling or shared restarts.
Known limits
This is static evidence from preserved real catalogs, not observed execution,
complete effective configuration, production-latency certification or removal
approval. A weak chosen witness does not prove every alternative path is weak.
Limits constrain annotations, not dependency membership or removal blockers.
Native catalogs contained no positive shared-caller path; positive shared/mixed/
bounded semantics remain adversarial automated contracts. Emulated guest readiness
recoveries and a non-recurring Splunk host-chrome startup error are documented in
the compatibility guide. Existing bundle-size and AppInspect advisories remain.
Customer content stays read-only; review workflow state is app-local only.
Install
Install or update with content_hygiene-1.12.0.tar.gz through Splunk Manage Apps.
No rescan or decision migration is required solely for path-context annotations.
Keep existing review records and follow the installation/administration guides.
SHA-256 (content_hygiene-1.12.0.tar.gz):
fcb7f01a9f1b41d5be4838c9f2df09ccf646fff61da3a616302bce73f3b32ccc
The archive contains only the installable app, with private reports, lab details,
credentials, development files and macOS metadata excluded.
Content Hygiene 1.11.0 — build 96
Caller-supported cross-dimension dependency paths
Content Hygiene 1.11.0 build 96 connects potential earlier-phase field producers
across host/source/sourcetype scopes only when one captured original-event query
independently selects both scopes in the same caller app. It never combines
unrelated searches or dashboard panels, joins on field names alone, or executes
customer SPL, PCRE, expressions, event sampling or lookup data for inference.
Supporting receipts retain both scopes plus exact caller, app, query and selector
locations. Native Splunk relationship details provide wrapping, filtering,
pagination and exact caller drill-through. Evidence reaches required-content,
removal-impact and CSV/JSON exports. Indexed visit/work/projection/byte ceilings
leave additional paths unverified. The stale attribution notice for newer
supported field-lineage snapshots is also fixed.
New complete scans use shared-static-v10; v1-v9 decisions, captured baselines,
notes, assignees and journals remain readable unchanged. Run a complete scan
after upgrading to record the expanded evidence explicitly.
Qualification
- 1,107 automated checks; lint, types, build and portable archive gates pass.
- AppInspect 4.3.1: zero errors/failures/future failures; three existing reviewed
advisories and one manifest skip. The main bundle retains its size advisory. - The same exact archive on Enterprise 10.2.5, 10.2.8 and 10.4.4: all seven pages
in Light/Dark, native/cache/source/schema comparisons, 108 accepted captures
and actual 34 JSON / 32 paired CSV comparisons. Minimum sampled active
text/link contrast 5.78:1; no measured horizontal overflow or graph-label overlap. - Original decisions/policy journals and native preferences preserved; no
customer-content writes, new review/policy saves, accounts or shared restarts.
The native catalogs contain no qualifying cross-dimension producer pair.
Positive semantics and supporting-caller presentation remain automated contracts,
not manufactured native witnesses. Unwitnessed joins, arbitrary PCRE, hidden
private/system/file layers, rewrites, actual field availability and runtime or
external/ad-hoc closure remain unverified. No absence or review stage authorizes
removal. Issue #8 remains complete against its documented finite static contract.
The owned minimum-version VM needed cold-start readiness recovery after its
startup wrapper expired, plus two browser-control batch recoveries. Complete
evidence was preserved and interrupted steps freshly verified; no frozen source,
application timeout or assertion changed. These are representative checks, not
zero-timeout or production-latency certification.
See the qualification details
and investigation guide.
Minimum support remains Enterprise 10.2.5+ on a standalone Linux x86-64 search
head with healthy KV Store; SHC, Windows coordinated writes and Splunk Cloud
remain outside qualification.
Assets
The archive excludes credentials, private reports, lab details, development/test
files, macOS metadata, extended tar headers and links. Assets are the exact
qualified archive and SHA256SUMS.
SHA-256 (content_hygiene-1.11.0.tar.gz):
7f92f36dece18441fe6fa5b906173344d12099237870f3507cc14ab0eae5ce26
Content Hygiene 1.10.0 — build 95
Content Hygiene 1.10.0 — build 95
Completes the bounded shared-static source-family contract in #8. Declared field
producers can now be investigated across compatible literal / * / ... scopes
and explicitly captured shared default fallbacks. Relationship context explains
both scopes and the static compatibility basis; it never claims observed events,
effective assignment or execution.
- Preserve phase order, exact namespace, local/private shadow checks and REPORT
output provenance. Source/sourcetype producer domains are case-sensitive; host
domains remain ASCII-insensitive. Unrelated dimensions are not guessed. - Keep direct exact-scope indexes and bound candidate/product-state work. Failed
matches consume visits; unsupported or exhausted paths stay partial. - Carry compact receipts into Explorer, required-content witnesses, impact and
CSV/JSON. Native Splunk controls, wrapping, keyboard access and Light/Dark modes
remain.shared-static-v9preserves decisions, journals and v1–v8 baselines. - Add a public acceptance ledger, investigation/troubleshooting guidance and
privacy-reviewed native diagnostics screenshots.
Qualification
The exact archive passes 1,065 checks, lint/style/types/build/portable-package
gates and AppInspect 4.3.1 pre-cert with zero errors/failures/future failures.
Three existing advisories (SplunkJS telemetry, Python migration and KV Store
configuration) and one manifest skip remain; the main bundle retains its size
advisory. Trusted-library validation uses the existing cache.
Enterprise 10.2.5, 10.2.8 and 10.4.4: fresh scans, native/cache/source replay,
all seven pages in Light/Dark, 106 accepted native captures, all 21 installed
app-owned page definitions matching the archive, and 34 actual JSON / 32 paired
CSV comparisons. All 11 schemas are valid; 21,118 current object/edge/scan records
and the preserved review baseline validate against applicable schemas. Minimum
sampled active text/link contrast is 5.78:1, with no measured horizontal overflow,
graph-label overlap or modal scope clipping. App browser errors are zero.
Existing decisions/policies and original appearance choices are preserved. No
new reviews, policies, accounts, source permissions, customer writes or shared
restarts were used. The owned minimum-version guest was gracefully stopped with
its disk/authentication preserved and task-owned listeners removed.
The native catalogs contain 46 literal producer links and 24 pattern declarations,
but no compatible captured pattern/default producer/caller positives or shared
defaults. Positive extension/collision/shadow/props-Library cases are automated
contracts, not manufactured native content. A native Indexes timeout correctly
retained the prior baseline before a successful full scan; unstable pagination
was rejected before stable capture. Cold emulated startup and initial/control
readiness waits recovered after fresh checks without changing production source
or timeouts. This is not zero-timeout or production-latency certification.
Boundaries
Customer content remains read-only. Completing #8 is not full effective-props,
private/system/file configuration, arbitrary PCRE/encoding, cross-dimension field
joins without independent witnesses, rewrites or runtime/external/ad-hoc closure.
Scheduled-versus-manual attribution remains separate. No missing edge, review
stage or coverage indicator authorizes removal.
See the acceptance ledger
and qualification boundaries.
Minimum runtime: Splunk Enterprise 10.2.5; standalone Linux x86-64 scope.
Archive SHA-256: 6b870533ec41d1fdba7844e68ba3fab588614cc242aca044aa20a469b329b1d2.
Content Hygiene 1.9.0 — build 94
Bounded props patterns and default fallbacks
Content Hygiene 1.9.0 build 94 extends the shared static dependency graph with
whole-name ASCII literal / * / ... pattern caller possibilities and explicitly
captured shared [default] fallbacks. No customer regex, SPL or EVAL is executed;
no events or lookup rows are sampled to manufacture a matching witness.
- Prefix/namespace indexes and separate candidate/character-work ceilings keep
unsupported and budget-limited paths explicitly unknown. - Reuse existing native stanza reads and budgets. Conditional same-attribute
ordering uses literal/pattern priority defaults of 100/0 and keeps defaults
below specific stanzas. Every lower/default/shadowed possible path remains. - Explorer, relationship details and Review Library explain support and limits
with native Splunk UI controls, wrapping, keyboard access and theme-safe links. - Facts flow through impact, required-content reports, CSV/JSON and freshness.
shared-static-v8preserves decisions, journals and v1-v7 baselines; expanded
evidence requires explicit re-review, not automatic reaffirmation.
Qualification
The exact archive passes 1,024 automated checks and lint/style/types/branding/
production-build/portable-package gates. Pre-cert AppInspect 4.3.1 reports zero
errors/failures/future failures, three existing reviewed advisories and one
missing-manifest skip, using the existing trusted-library cache. Bundle-size
advisories remain; this is not Cloud certification.
Fresh native/cache/schema and caller-edge replay, all seven pages in Light/Dark,
103 accepted captures and 29 JSON / 27 paired CSV downloads cover Enterprise
10.2.5, 10.2.8 and 10.4.4. Sampled active text/links have contrast at least 5.78:1,
with no measured horizontal overflow, graph-label overlap or dialog overflow.
Original decisions, baselines, journals and appearance choices are preserved.
Customer content remains read-only; no new reviews/policies/accounts or shared
restarts were used for qualification.
Honest limits
Native catalogs contain 24 supported pattern declarations but no compatible
captured pattern callers or shared default declarations. Positive pattern/default
ordering, collisions/shadowing and props-specific Library cases are adversarial
automated contracts, not fabricated native records. Full effective props,
arbitrary PCRE/modifiers, Windows/encoded/non-ASCII scopes, private/system/file
layers, cross-pattern producer joins and runtime/dynamic/external consumers remain
unverified. #8 stays open. No missing edge or conditional order approves removal.
The emulated minimum-runtime startup wrapper and three Dark initial-navigation
readiness waits expired. Fresh native readiness, exact page/root checks and full
assertions subsequently passed on the same disk/authentication, with no source,
production timeout or assertion changed. No app browser errors were recorded.
These bounded desktop checks are not a zero-timeout emulated run, production
timing benchmark or comprehensive accessibility certification.
See the investigation guide
and qualification boundaries.
Install and integrity
Requires Splunk Enterprise 10.2.5+ on a supported standalone Linux x86-64 search
head with healthy KV Store and bundled Python. Install the attached archive;
preserve review KV Store state during upgrades. Search-head clustering, Windows
coordinated review writes and Splunk Cloud remain outside support.
content_hygiene-1.9.0.tar.gz
SHA-256: adb4ddfd6989f55136a47d039aa567ecb8ab5ceed9bffa344c4909b548aaf399
The portable archive excludes macOS metadata, development files, reports, lab
details and credentials. Public screenshots contain reviewed built-in metadata;
raw validation reports remain private.
Content Hygiene 1.8.0 — build 93
Conditional cross-stanza props investigations
Content Hygiene 1.8.0 build 93 adds conditional ordering of the same exact props
attribute across captured shared literal stanzas for one supported original-event
selector conjunction. Category ordering precedes within-category native numeric
priority and ASCII ties. Separate queries and Boolean branches are never combined.
- Explain highest/lower captured precedence versus unverified context, retaining
every potential dependency path and lower/shadowed removal fallback. - Reuse existing scan-time native reads; bound work, time, candidates, rows and
bytes. No customer SPL/PCRE, workflow or external destination is executed. - Add wrapping, searchable, filtered and paginated details in Explorer, Library
and relationship dialogs, with keyboard caller drill-through and native themed,
underlined links in both appearance modes. - Carry compact facts/provenance through freshness, impact, required-content,
scan reports and CSV/JSON.shared-static-v7preserves decisions, journals and
v1-v6 captured baselines; a scan never silently reaffirms a decision.
Qualification
The exact archive passes 968 automated checks, lint/style/types/build/branding,
portable metadata-free packaging and pre-cert AppInspect 4.3.1 with zero errors,
failures or future failures. Three existing advisories and one manifest-related
skip remain reviewed; existing bundle-size advisories remain.
Enterprise 10.2.5, 10.2.8 and 10.4.4 have fresh bounded/full scans, independent
native/cache/schema comparisons, all seven pages in Light/Dark, 101 accepted
captures, keyboard caller/context checks and 29 actual JSON downloads with 27
paired CSV comparisons. Sampled active text/link contrast is at least 5.78:1.
Original decisions, baselines, review/policy journals and appearance choices are
preserved. No new reviews, policies, accounts or customer objects were saved.
Shared installations were not restarted. Private validation reports, lab details
and credentials are excluded from the public source and assets.
Scope and limits
Native catalogs contain no positive exact host/source or priority/ASCII collision
witness. Positive collision, shadowing and props-specific Library cases have
adversarial automated contracts, not manufactured native records. The emulated
minimum runtime needed cold-load/readiness recovery; this is not production-scale,
comprehensive accessibility or Cloud certification.
This is not full effective props or execution evidence. Patterns/defaults,
private/system/file layers, sourcetype rewrites, field availability, runtime
activation and dynamic/external/ad-hoc consumer closure remain unverified.
No missing edge or ordering result establishes safe removal. #8 remains open.
See the investigation guide,
coverage matrix
and qualification boundaries.
Minimum support remains Enterprise 10.2.5 on a standalone Linux x86-64 search head
with healthy KV Store. Upgrade using the attached archive; package-policy checks
exclude macOS metadata and development/private files.
SHA-256 (content_hygiene-1.8.0.tar.gz):
d83cf4fb09b9190ad384f782cf10353422d756a36bf76d2865bf51228175ccd8
Content Hygiene 1.7.0 — build 89
Content Hygiene 1.7.0 — build 89
This release adds a bounded, read-only shared props precedence and inheritance
cross-check. It compares native rule-level ownership/export and the merged value
of the same exact stanza/attribute with the captured shared definition.
- Distinguish local merged matches, inherited merged matches, shadowed bindings
and unverified contexts, with specific reasons and a returned-binding drill-down. - Inspect accessible, wrapping, searchable and paginated details in Dependency
Explorer, Review Library and relationship dialogs. Retain the same evidence in
review freshness, removal impact, required-content reports and CSV/JSON. - Keep shadowed paths as possible removal fallbacks. Never equate a configured
binding with execution or interpret an unassessed context as absence. - Bound scan-time reads, response bytes, candidate work, fan-out and elapsed time.
Context browsing does not issue additional native reads or run customer searches. - Preserve existing decisions, notes, assignees, baselines and journals. New
shared-static-v6evidence can request explicit re-review; scans do not reaffirm
or rewrite saved choices. Earlier v1–v5 baselines remain readable.
Qualification
The exact attached archive passes 907 automated checks, lint/style/types,
branding, production build and portable archive verification. Pre-cert
AppInspect 4.3.1 reports zero errors/failures, with three existing reviewed
advisories and one manifest-based skip; this is not Splunk Cloud certification.
Native qualification covers Enterprise 10.2.5, 10.2.8 and 10.4.4: fresh bounded/full
scans, independent GET-only catalog/cache replay, published schemas, all seven
pages in Light/Dark, focused props investigations and 29 JSON plus 27 paired CSV
downloads. Existing decisions and policy history are unchanged; original
appearance preferences are restored. No customer objects, review decisions or
policies were saved for this slice. Shadowing collisions and props-specific
Library positives have automated contracts, not manufactured native witnesses.
Compatibility and limits
Minimum: Splunk Enterprise 10.2.5, standalone Linux x86-64 search head with KV Store.
Run a new complete scan after upgrading to collect shared props observations.
No new REST exposures, roles, source-write permissions or shared-host restart
were required for qualification.
This is not full effective-props evaluation or byte-for-byte source-file origin.
Private/system inheritance, file layers, cross-stanza source/host/sourcetype
priority, runtime activation/field availability and dynamic/external/ad-hoc closure
remain unverified. Follow-up #8 stays open; missing relationships never prove
safe removal. Customer content remains read-only.
See the investigation guide,
coverage matrix
and qualification boundaries.
Integrity
content_hygiene-1.7.0.tar.gz
SHA-256: 7f7f900ea62a730170ba24227b698f970335132f3830db3f6fc54ec0ba13826e
The package excludes credentials, private validation reports, lab details,
development files and macOS metadata. Verify against the attached SHA256SUMS.