[Threat Intel 3/11] HTML and article content parsing by stephmilovic · Pull Request #287199 · elastic/kibana

stephmilovic · 2026-08-25T22:30:48Z

DO NOT REVIEW! Considering alternative approach

Turns fetched feed content into the content shape a threat report stores. Nothing imports it yet.

text parses markup through JSDOM, removes non-content and hidden subtrees, collapses whitespace, truncates to the stored-body cap, and builds the report content object. It also classifies section headers, which IOC extraction later uses to decide whether a value appeared in an indicators table or in prose. Inline visibility is evaluated with css-tree, so malformed HTML and CSS follow parser behavior instead of custom lexical scanning.

extract_article uses Mozilla Readability to pick the article body out of a full page. canonicalize_url normalizes URLs for dedup, and severity maps a level to the numeric score the report stores.

One behaviour worth a look: buildReportContent falls back to the title when the body is empty. Every enrichment route requires a non-empty text, so a report stored with no body can never be enriched, and it keeps getting picked up by the pending scan and failing. Title-only entries are common in feeds that carry a headline and a link, and a headline is thin but real input. The fallback lives here rather than in each adapter so all six get it.

This layer sits below extraction and above the adapters, which is why it is its own directory rather than part of adapters: IOC extraction depends on it, and the adapters depend on both.

Where this sits

#	PR	Depends on	Status
1	#287197 workflow gate correction + fixtures	nothing	✅ merged
2	#287198 contracts, constants, shared libs	nothing	✅ merged
3	#287199 content parsing ← this PR	#287198	closed
4	#287200 SSRF-guarded HTTP client	#287198	✅ merged
5	#287201 index templates, seeding, inference features	#287198, #287200	✅ merged
6	#287202 indicator alias	#287198, #287201	✅ merged
7	#287203 IOC extraction	#287198, #287200	✅ merged
8	#287204 LLM services, routes, source catalog API	#287198, #287200, #287202, #287203	👀 ready
9	#287205 RSS adapter	#287198, #287200, #287203, #287204	draft
10	#287206 remaining adapters, dispatcher, fetch_source step	#287198, #287200, #287203, #287205	draft
11	#287207 promote/scrub tasks and plugin wiring	#287198–#287206	draft
12	#287479 generator fixture article URLs	nothing	✅ merged

#287479 came first in the merge train and is already on main. The numbered series (#287197–#287207) stacks on top in dependency order. #287199 (content parsing) was closed; IOC extraction ships its own section-header classifier.

New third-party dependencies

`css-tree@3.2.1`

Purpose: Parses inline CSS declaration lists and values used to decide whether extracted HTML is hidden by display, visibility, all, priorities, escapes, or supported var() fallbacks.
Justification: Visibility is part of the extraction security boundary. A standards-based CSS parser gives us declaration recovery, identifier decoding, grammar validation, and cascade data without maintaining another CSS scanner.
Alternatives explored: The earlier implementation split and scanned declarations itself. Review found malformed-input bypasses in that approach, so it was removed. JSDOM handles the HTML DOM, but its CSSOM does not provide the declaration AST and explicit recovery behavior this evaluator needs.
Existing dependencies: Kibana has CSS tooling for build and transformation paths, and css-tree already appears transitively in the lockfile. This code imports it at server runtime, so it is declared directly and pinned to the repository-resolved version.

`@mozilla/readability@0.6.0`

Purpose: Selects the primary article content from a normalized document before the threat report text is serialized.
Justification: Article selection is a mature heuristic problem. Readability replaces the custom candidate scoring with a maintained implementation used for this exact job, while our code keeps only the threat-intelligence-specific normalization and fallback policy.
Alternatives explored: The previous implementation scored candidate elements itself. That added bespoke heuristics and edge cases, so it was removed. Keeping the whole normalized body is the fallback, but using it as the primary path would retain navigation, teasers, and unrelated page content.
Existing dependencies: JSDOM already parses HTML, but it does not identify the main article. No existing Kibana utility in this runtime path provides Readability-equivalent article selection.

`@types/css-tree@2.3.10`

Purpose: Supplies TypeScript declarations for the css-tree runtime API used by the inline-style evaluator.
Justification: css-tree does not ship the required TypeScript declarations. The direct runtime import needs checked types in Kibana's TypeScript build.
Alternatives explored: Local ambient declarations or broad casts would duplicate the public API and hide library drift. The DefinitelyTyped package keeps that surface checked and reviewable.
Existing dependencies: TypeScript and the existing DOM typings do not describe css-tree; there is no overlapping type package. This is compile-time only.

Inline CSS evaluation intentionally covers declarations and custom properties on the element itself. It does not load external stylesheets or inherit custom properties from ancestors. Unsupported CSS stays visible rather than being guessed as hidden.

To test

node scripts/jest x-pack/solutions/security/plugins/security_solution/server/threat_intel/content
node scripts/type_check --project x-pack/solutions/security/plugins/security_solution/tsconfig.json

The content suites and scoped type check should pass.

PR developed with Claude Code + Claude Opus 5

Third slice of the threat intel supply pipeline. Turns fetched feed content into the `content` shape a threat report stores. Nothing imports it yet and `threatIntelSupplyEnabled` defaults off, so it is inert. `text` strips markup, collapses whitespace, truncates to the stored-body cap, and builds the report `content` object. It also classifies section headers, which IOC extraction later uses to decide whether a value appeared in an indicators table or in prose. Script and style stripping matches end tags with attributes and arbitrary whitespace, since a regex that only handles `</script>` is not a sanitizer. `extract_article` picks the article body out of a full page, `canonicalize_url` normalizes URLs for dedup, and `severity` maps a level to the numeric score the report stores. One behaviour worth calling out: `buildReportContent` falls back to the title when the body is empty. Every enrichment route requires a non-empty text, so a report stored with no body can never be enriched, and it keeps getting picked up by the pending scan and failing. Title-only entries are common in feeds that carry a headline and a link, and a headline is thin but real input. The fallback lives here rather than in each adapter so all six get it. This layer sits below extraction and above the adapters, which is why it is its own directory rather than part of `adapters`: IOC extraction depends on it, and the adapters depend on both. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

infra-vault-gh-plugin-prod · 2026-08-26T16:22:57Z

Pinging @elastic/security-threat-hunting (Team:Threat Hunting)

infra-vault-gh-plugin-prod · 2026-08-26T16:22:59Z

Pinging @elastic/security-solution (Team: SecuritySolution)

kibanamachine

Libra found 4 issues.

Generated by Libra

Eight review findings. The article container was picked by first-match-of-first-selector, and two problems compounded: ARTICLE_SELECTORS puts `article` ahead of `main`, so any <article> beat a <main> holding the real report, and `.first()` then took the earliest of those. A page with an <article> teaser card above the body returned the teaser and every IOC in the report was missed. Candidates are now gathered across all selectors and the one with the most text wins, with selector order as a tie-break. Headings were classified before entity decoding, so `Indicators of Compromise` read as prose and its anchor hrefs were dropped rather than lifted. Every heading reset the section kind regardless of depth, so `<h2>Indicators of Compromise</h2><h3>Domains</h3>` fell back to prose at the subsection that actually holds the indicators. A deeper unclassified heading now inherits its parent section; a same-or-higher heading, or any explicitly classified one, still ends it. The href lift only matched quoted attributes. Unquoted href is valid HTML, and the generic tag stripper then removed the attribute, so an href-only IOC vanished. `ref` was stripped as a tracking param, but GitHub and GitLab use it for branch and commit refs and threat reports link to repos constantly, so it produced wrong canonical keys for exactly the URLs this pipeline sees most. Only unambiguous campaign values are stripped now. `www.com` is a registered domain and `slice(4)` turned it into the bare TLD `com`. Stripping now requires a dot after position 4. The three parser entry points took unbounded input from fetched pages inside a task worker. They cap at 10MB, matching the `body_html` bound the report API enforces. Truncating rather than throwing keeps a fat page degraded instead of failed. The title fallback for an empty body was silent, so the document was indistinguishable from one that genuinely repeats its title and enrichment paid to run inference over the same string twice. It sets `body_is_title_fallback` now. Also added the requested test coverage for `section_headers` and `severity`, and dropped an underscore prefix on a non-exported helper. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

stephmilovic · 2026-08-26T17:49:35Z

Thanks @jonwalstedt, all eight addressed, details in the threads.

Two worth surfacing here rather than leaving buried:

The container-selection fix took two passes. My first attempt only picked the largest match within a selector, which misses the real problem: ARTICLE_SELECTORS puts article ahead of main, so the loop returned the teaser before main was ever considered. Candidates are now compared across all selectors.

The body_is_title_fallback field you asked for needs a mapping entry too, since the reports index is dynamic: strict. That side is in #287201, so the two should land together.

kibanamachine

Libra found 3 issues.

Generated by Libra

…hor boundaries Three review findings, two of them gaps in fixes from the previous round. Container candidates were scored on raw text before chrome removal, so a teaser carrying a large inline script or style could outweigh the real report, win selection, and then be stripped to almost nothing. That drops the report and every IOC in it. Each candidate is now scored on a clone with the same chrome removed that the returned container gets. This is a gap in the cross-selector fix from the last round, which changed which candidates were compared but not how they were measured. `truncate` sliced to `maxLength` and then appended the ellipsis, so every truncated value came back one character over the cap. Callers pass the cap to satisfy a downstream length check, so being one over defeats the point. The ellipsis is now reserved inside the cap, with a zero or negative cap returning empty rather than a bare ellipsis. The anchor pattern had no attribute boundary before `href`, so the greedy prefix could run past `data-href="..."` and lift the tracker instead of the link. In an IOC section that both loses the real indicator and invents a false one. Requiring whitespace before `href` fixes it, and also allows whitespace around the equals sign, which is valid HTML. This one was made more reachable by accepting unquoted values in the previous round. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

`NAMED_ENTITIES` was an object literal, so it inherited `Object.prototype` and a lookup for `constructor`, `toString`, `valueOf`, `hasOwnProperty`, or `__proto__` returned a real value. The `!== undefined` guard then accepted it, so feed HTML containing `&constructor;` decoded to `function Object() { [native code] }` and `&__proto__;` to `[object Object]`. That text lands in `content.body_text`, which is what the LLM enrichment stages read and what IOC extraction runs over, so untrusted feed markup could inject arbitrary filler into a stored report with a nine-character string. It is a `Map` now, which has no inherited keys. Found while re-reading the diff rather than reported. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

stephmilovic · 2026-08-26T18:31:51Z

All three addressed, details in the threads. Two of them were gaps in fixes I made in the previous round, which is worth naming: the candidate-scoring change compared the right candidates but still measured them wrong, and accepting unquoted hrefs widened what the greedy prefix could swallow, making the data-href capture likelier than before.

I also did another pass over the diff and found one more, unreported:

NAMED_ENTITIES was an object literal, so it inherited Object.prototype. A lookup for constructor, toString, valueOf, hasOwnProperty, or __proto__ returned a real value, the !== undefined guard accepted it, and feed HTML containing &constructor; decoded to function Object() { [native code] }. That text goes into content.body_text, which is what the LLM stages read and what IOC extraction runs over, so nine characters of untrusted markup could inject arbitrary filler into a stored report. It is a Map now, with tests for each prototype member name plus the real entities.

Mirror the elasticsearch-controller changes. Both _search_ai_lake_analyst and _search_ai_lake_soc_manager can run Detection Engine Indicator Match rules, which need read access to the per-space threat intel aliases.

Rules only need read access. Drop view_index_metadata for now, can be added in a follow-up if needed.

github-actions

Reviewed the content-parsing layer. The source modules are well-structured and thoroughly tested (including the Chromium fuzz oracle), and I found no concrete correctness, security, or coverage defects in them. One inline finding on accidentally committed local Bazel artifacts.

Generated by Claude Reviewer for #287199 · opus · 106.2 AIC · ⌖ 14.3 AIC · ⊞ 5.1K

Adapters in the RSS/STIX/TAXII (elastic#287205) and vendor_api (elastic#287206) slices import collapseWhitespace from content/text.ts. The content-parsing rewrite made it a private const, which breaks those imports once this PR merges. It's a plain string utility with real external callers, not part of the parsing-architecture change, so restore the export.

…ction_headers helper The content-parsing refactor (elastic#287199) internalized IOC_HEADER_TERMS, TERMINATOR_HEADER_TERMS, TERMINATOR_PREFIXES, and normalizeHeader behind a single classifyHeader() entry point that also added plural-aware and prefix-boundary-aware matching this file's own isIocHeader/isTerminatorHeader lacked. Those raw exports no longer exist, so this file's import breaks once elastic#287199 merges. Delegate to classifyHeader directly instead of re-implementing the same classification locally.

…el-serverless-roles PR)

kibanamachine · 2026-08-28T15:04:40Z

💛 Build succeeded, but was flaky

Buildkite Build
Commit: c08aa64
Build duration: 62 mins

Failed CI Steps

Metrics [docs]

✅ unchanged

Test Failures

[job] [logs] Scout Lane #77 - serverless-search / agent_builder / local-serverless-search - Agent Builder — concurrent conversation writes API - keeps both rounds when two clients converse with the same conversation at once
[job] [logs] Scout Lane #21 - stateful-classic / default / local-stateful-classic - getCertificates - returns an empty result when no monitors are configured
[job] [logs] Jest Tests #16 / Severity form field closes the bulk actions correctly

History

💔 Build #493117 failed 9232274
💔 Build #493107 failed 5708ee8
💔 Build #493093 failed 51acddb
💚 Build #493069 succeeded 7332792
💔 Build #493039 failed 1684cb9

infra-vault-gh-plugin-prod · 2026-08-28T16:10:38Z

🤖 Jobs for this PR can be triggered through checkboxes. 🚧

ℹ️ To trigger the CI, please tick the checkbox below 👇

Click to trigger kibana-pull-request for this PR!
Click to trigger kibana-deploy-project-from-pr for this PR!
Click to trigger kibana-deploy-cloud-from-pr for this PR!
Click to trigger kibana-entity-store-performance-from-pr for this PR!
Click to trigger kibana-storybooks-from-pr for this PR!

…ction_headers helper The content-parsing refactor (elastic#287199) internalized IOC_HEADER_TERMS, TERMINATOR_HEADER_TERMS, TERMINATOR_PREFIXES, and normalizeHeader behind a single classifyHeader() entry point that also added plural-aware and prefix-boundary-aware matching this file's own isIocHeader/isTerminatorHeader lacked. Those raw exports no longer exist, so this file's import breaks once elastic#287199 merges. Delegate to classifyHeader directly instead of re-implementing the same classification locally.

jonwalstedt · 2026-08-28T20:00:28Z

Closing per the TI MVP scope-narrowing decision. Arbitrary web-page / HTML article ingestion (mozilla/readability, css-tree, htmlToStructured, extractArticleHtml) is out of scope for the MVP, so this PR's content/ module is no longer needed as a standalone diff.

The small generic helpers that downstream PRs actually used have been transplanted onto their owning PRs (imports rewritten, no shared ancestry):

classifyHeader → services/section_headers.ts ([Threat Intel 7/11] IOC extraction with a precision pass and confidence tiering #287203)
severityScore / severity defaults → services/severity.ts, report-content helpers → services/report_content.ts ([Threat Intel 8/11] LLM enrichment services, routes, and source catalog API #287204)
RSS-local htmlFragmentToText (bounded plain-text conversion, the one narrow markup boundary we keep) → adapter-local in [Threat Intel 9/11] RSS adapter with bounded structured-text serialization #287205
canonicalizeUrl → adapter-local in [Threat Intel 10/11] Remaining adapters, dispatcher, and fetch_source step #287206

Nothing from this branch is lost — the pieces that remain in scope now live with their consumers. Closing rather than merging.

…ction_headers helper The content-parsing refactor (elastic#287199) internalized IOC_HEADER_TERMS, TERMINATOR_HEADER_TERMS, TERMINATOR_PREFIXES, and normalizeHeader behind a single classifyHeader() entry point that also added plural-aware and prefix-boundary-aware matching this file's own isIocHeader/isTerminatorHeader lacked. Those raw exports no longer exist, so this file's import breaks once elastic#287199 merges. Delegate to classifyHeader directly instead of re-implementing the same classification locally.

…og API (#287204) ## Summary Adds LLM enrichment services and Threat Intelligence HTTP routes, including the fixed-catalog source list and enable/disable API. Manual report creation and provenance URLs go through the shared HTTP/HTTPS normalizer. Also wires the routes, inference features, and one-time bootstrap into `plugin.ts`, gated by `threatIntelSupplyEnabled`. This wiring is interim: it exists so the enrichment eval suite (see Evaluation below) has routes to call ahead of the full pipeline wiring. `#287207` replaces it with a proper `wiring.ts` module; that PR's description has a rebase note covering the `plugin.ts` conflict resolution. ## Where this sits | # | PR | Depends on | Status | |---|---|---|---| | 1 | #287197 workflow gate correction + fixtures | nothing | ✅ merged | | 2 | #287198 contracts, constants, shared libs | nothing | ✅ merged | | 3 | #287199 content parsing | #287198 | closed | | 4 | #287200 SSRF-guarded HTTP client | #287198 | ✅ merged | | 5 | #287201 index templates, seeding, inference features | #287198, #287200 | ✅ merged | | 6 | #287202 indicator alias | #287198, #287201 | ✅ merged | | 7 | #287203 IOC extraction | #287198, #287200 | ✅ merged | | 8 | #287204 LLM services, routes, source catalog API **← this PR** | #287198, #287200, #287202, #287203 | 👀 ready | | 9 | #287205 RSS adapter | #287198, #287200, #287203, #287204 | draft | | 10 | #287206 remaining adapters, dispatcher, fetch_source step | #287198, #287200, #287203, #287205 | draft | | 11 | #287207 promote/scrub tasks and plugin wiring | #287198–#287206 | draft | | 12 | #287479 generator fixture article URLs | nothing | ✅ merged | | 13 | #289343 Scout API tests for the deterministic routes | #287204 | draft | | 14 | #289345 enrichment eval suite | #287204 | draft | #287479 came first in the merge train and is already on `main`. The numbered series (#287197–#287207) stacks on top in dependency order. #287199 (content parsing) was closed; IOC extraction ships its own section-header classifier. ## Scope review follow-up - Shared `normalizeProvenanceUrl` helper (HTTP/HTTPS only, strips credentials, bounded length). - Source create/delete routes removed. Updates may change only `enabled`. - Lists and mutations reject catalog IDs outside the approved set. - Ensures the per-space indicator alias on `list_sources` (idempotent). ## Evaluation A `@kbn/evals` suite exercises all four enrichment stages against the BlackHat demo pack article text, posting each input directly to the internal route and scoring the structured response. It runs as a tracked baseline (scores recorded, no build-failing thresholds yet). Latest full run, EIS Claude Sonnet 4.6 as model and judge, 1 repetition, 20 examples: | Stage | Evaluator | Score | |---|---|---| | assess_relevance | IsIntelligenceMatch | 1.00 | | assess_relevance | RelevanceShapeValid | 1.00 | | classify_severity | SeverityExactMatch | 0.83 | | classify_severity | SeverityWithinOneLevel | 1.00 | | enrich_taxonomy | CategoryRecall | 0.80 | | enrich_taxonomy | RegionRecall | 1.00 | | extract_diamond | DiamondNoIocLeak | 1.00 | | extract_diamond | DiamondSignalCount | 1.00 | | extract_diamond | criteria (LLM judge, majority vote) | 1.00 | The suite ships in #289345, a separate eval-only PR that stacks on this one (it needs the routes and the flag-gated wiring added here). The four deterministic routes are covered separately by Scout API tests in #289343. Any prompt or calibration fix the evals surface lands by amending this PR, not the eval PR. ## To test ``` node scripts/jest --config x-pack/solutions/security/plugins/security_solution/server/threat_intel/jest.config.js routes/list_sources.test.ts services/provenance_url.test.ts ``` _PR developed with Cursor + Auto_ --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> Co-authored-by: Jon Wålstedt <jon.walstedt@elastic.co> Co-authored-by: Cursor <cursoragent@cursor.com>

…d the flag (#287207) ## Summary Wires Threat Intelligence behind `threatIntelSupplyEnabled`: bootstrap, managed workflow installation, promote/scrub tasks, and route registration. Promote mirrors extracted IOCs into `.threat-intel-indicators` for Indicator Match rules. ## Where this sits | # | PR | Depends on | Status | |---|---|---|---| | 1 | #287197 workflow gate correction + fixtures | nothing | ✅ merged | | 2 | #287198 contracts, constants, shared libs | nothing | ✅ merged | | 3 | #287199 content parsing | #287198 | closed | | 4 | #287200 SSRF-guarded HTTP client | #287198 | ✅ merged | | 5 | #287201 index templates, seeding, inference features | #287198, #287200 | ✅ merged | | 6 | #287202 indicator alias | #287198, #287201 | ✅ merged | | 7 | #287203 IOC extraction | #287198, #287200 | ✅ merged | | 8 | #287204 LLM services, routes, source catalog API | #287198, #287200, #287202, #287203 | ✅ merged | | 9 | #287205 RSS adapter | #287198, #287200, #287203, #287204 | ✅ merged | | 10 | #287206 remaining adapters, dispatcher, fetch_source step | #287198, #287200, #287203, #287205 | ✅ merged | | 11 | #287207 promote/scrub tasks and plugin wiring **← this PR** | #287198–#287206 | ready for review | | 12 | #287479 generator fixture article URLs | nothing | ✅ merged | | 13 | #289343 Scout API tests for the deterministic routes | #287204 | ready for review | | 14 | #289345 enrichment eval suite | #287204 | ready for review | | 15 | #290144 read APIs, readiness, space-keyed attribution | #287207 | draft | #287479 came first in the merge train and is already on `main`. The numbered series (#287197–#287205) is merged; #287206 and #287207 now sit directly on `main`. #287199 (content parsing) was closed; IOC extraction ships its own section-header classifier. ## Scope review follow-up - Installs the managed workflows with the right space topology: `attribute_alerts_to_reports` installs per space (it queries `.alerts-security.alerts-*` and writes per-space hit totals, so a global install would clobber every space's totals onto one shared doc), while `ingest_threat_feeds` and `enrich_threat_report` install once globally. Alert analysis and threat intel share a single `ready()` call so neither install set is dropped, and spaces created after boot are reconciled on the promote task. All three ship `enabled: false`, and enrich routes its HTTP calls through a fixed space (`default`) instead of the install-time `workflow.spaceId` (which is `'*'` globally). - Ensures the default-space indicator alias on start. - Chunks promotion bulk writes and treats HTTP 408/500 as retryable bulk failures. - Sanitizes provenance and extracted IOC reference URLs during promotion. - Documents direct-index cross-space isolation as the blocker to enabling the feature. ## Bootstrap fix folded in from #289343 `seed_default_sources.ts` sorted the legacy-source disable scan on `_id`, which Elasticsearch rejects with `illegal_argument_exception: Fielddata access on the _id field is disallowed` unless `indices.id_field_data.enabled` is set, so it failed bootstrap on a stock cluster. Found this while getting #289343's Scout suite green and moved the fix here since this PR is the one wiring up bootstrap and already needs `security-threat-hunting` review, so it isn't adding a reviewer #289343 wouldn't otherwise need. ## Product boundary `threatIntelSupplyEnabled` defaults to false. Do not enable until direct-index cross-space isolation is hardened or the administrator trust model is explicitly accepted. ## To test ``` node scripts/jest --config x-pack/solutions/security/plugins/security_solution/server/threat_intel/jest.config.js wiring.test.ts tasks/promote_threat_indicators.test.ts node scripts/jest x-pack/solutions/security/plugins/security_solution/server/workflows/ node scripts/jest src/platform/packages/shared/kbn-workflows/managed/definitions/threat_intel/ ``` _PR developed with Cursor + Auto + Sonnet 5 + Opus 4.8_ --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> Co-authored-by: Jon Wålstedt <jon.walstedt@elastic.co> Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com>

jonwalstedt reviewed Aug 26, 2026

View reviewed changes

stephmilovic force-pushed the threat-intel-3-content branch from b19cfe2 to 38baf8c Compare August 26, 2026 16:22

stephmilovic marked this pull request as ready for review August 26, 2026 16:22

stephmilovic requested a review from a team as a code owner August 26, 2026 16:22

stephmilovic added the reviewer:libra PR review with Libra. This disables Claude and Scout reviewers label Aug 26, 2026

kibanamachine reviewed Aug 26, 2026

View reviewed changes

stephmilovic and others added 2 commits August 26, 2026 12:16

Changes from security: 3rd-party dependencies

2501b0c

elastic-vault-github-plugin-prod Bot requested a review from a team as a code owner August 27, 2026 23:55

elastic-vault-github-plugin-prod Bot requested a review from jeramysoucy August 27, 2026 23:55

stephmilovic removed the reviewer:libra PR review with Libra. This disables Claude and Scout reviewers label Aug 28, 2026

stephmilovic added 3 commits August 27, 2026 19:18

fix(threat-intel): use standards-based content parsing

9232274

fix(deps): assign threat intelligence dependency owners

0571b2a

Grant threat intel indicator aliases to AI SOC engine roles

fb6cf7b

Mirror the elasticsearch-controller changes. Both _search_ai_lake_analyst and _search_ai_lake_soc_manager can run Detection Engine Indicator Match rules, which need read access to the per-space threat intel aliases.

stephmilovic requested a review from a team as a code owner August 28, 2026 12:51

Remove view_index_metadata from threat intel indicator grants

5f9b3b5

Rules only need read access. Drop view_index_metadata for now, can be added in a follow-up if needed.

github-actions Bot reviewed Aug 28, 2026

View reviewed changes

Comment thread bazel-bin Outdated

stephmilovic added 2 commits August 28, 2026 07:33

Revert threat intel AI SOC engine role changes (belongs on threat-int…

1abe499

…el-serverless-roles PR)

chore: remove generated Bazel artifacts

c08aa64

stephmilovic removed request for a team and jeramysoucy August 28, 2026 14:13

stephmilovic marked this pull request as draft August 28, 2026 16:10

jonwalstedt closed this Aug 28, 2026

stephmilovic mentioned this pull request Sep 1, 2026

[Threat Intel] Move the indicator index out of the .kibana namespace #287479

Merged

This was referenced Sep 4, 2026

[Threat Intel] Scout API tests for the deterministic enrichment routes #289343

Merged

[Threat Intel] Enrichment eval suite for the LLM-backed routes #289345

Merged

Uh oh!

Conversation

stephmilovic commented Aug 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Where this sits

New third-party dependencies

css-tree@3.2.1

@mozilla/readability@0.6.0

@types/css-tree@2.3.10

To test

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

infra-vault-gh-plugin-prod Bot commented Aug 26, 2026

Uh oh!

infra-vault-gh-plugin-prod Bot commented Aug 26, 2026

Uh oh!

kibanamachine left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

stephmilovic commented Aug 26, 2026

Uh oh!

kibanamachine left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

stephmilovic commented Aug 26, 2026

Uh oh!

github-actions Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

kibanamachine commented Aug 28, 2026

💛 Build succeeded, but was flaky

Failed CI Steps

Metrics [docs]

Test Failures

History

Uh oh!

infra-vault-gh-plugin-prod Bot commented Aug 28, 2026

Uh oh!

jonwalstedt commented Aug 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

stephmilovic commented Aug 25, 2026 •

edited

Loading

`css-tree@3.2.1`

`@mozilla/readability@0.6.0`

`@types/css-tree@2.3.10`