agent-bober 0.17.1 → 0.19.0

package/skills/bober.seo-generic/SKILL.md

@@ -0,0 +1,173 @@
+---
+name: bober.seo-generic
+description: "Generic SEO/GEO signature library shared across every workflow-specific SEO skill. Not a workflow skill -- a data file of discrete, cited SEO/GEO tactics read by SeoPlaybookParser. Encodes VERIFIED 2025-2026 AI-visibility correlation research (Ahrefs, Semrush, Profound), ranking-mechanics leak findings, and the never-encode policy boundary."
+---
+
+# bober.seo-generic — Generic SEO/GEO Signature Library
+
+This skill is a **signature-library** file, not a workflow skill. It is read (as raw
+markdown text) by `SeoPlaybookParser.parse()` (`src/seo/parser.ts`) and turned into
+typed `SeoSignature[]` records used by the SEO agent team's retriever
+(`SeoPlaybookRetriever`) as the always-included generic floor. Every surviving
+signature below cites its primary source; a block with no `PrimarySourceUrl` or
+tagged `PolicyClass: never-encode` is DROPPED by the parser at load time — see
+`.bober/research/research-20260715-ultimate-seo-agents-skills-research.md` §1/§6 for
+the underlying evidence and the two refuted claims that are deliberately NOT encoded
+anywhere in this file (ChatGPT-Wikipedia dominance, Reddit-leading-AI-Overviews).
+
+## Signature Block Format
+
+Each signature is a level-3 heading (three `#` characters, a space, then the
+`playbookId`) followed by labelled fields. This file and `SeoPlaybookParser` are one
+executable spec — keep them in sync. Unlike the security signature library, SEO
+blocks carry **no code fences** — `Tactic` replaces the unsafe/safe example pair.
+
+Fields per block:
+- The heading text itself is the `playbookId` (must be non-empty, or the block is dropped).
+- `- **Title:** <human-readable title>` (required — missing drops the block)
+- `- **Workflows:** comma, separated, SeoWorkflow, members` (soft field; invalid members are filtered out, default `[]`)
+- `- **Tactic:** <the recommended action>` (soft field, defaults to `""` if absent)
+- `- **Invariant:** <the evidence-backed claim this signature encodes>` (soft field, defaults to `""` if absent)
+- `- **PrimarySourceUrl:** <REQUIRED citation URL>` — **a block missing or with an empty value here is DROPPED** (no-uncited-claim rule; the whole point of this format)
+- `- **PolicyClass:** auto-safe|human-approve|never-encode` — **`never-encode` is DROPPED**; any other invalid value is also DROPPED
+- `- **EvidenceGrade:** verified|primary-unverified|single-source` (soft field, defaults to `single-source` if absent/invalid)
+- `- **LiveWeightStatus:** live-corroborated|documented-only|unknown` (soft field, defaults to `unknown` if absent/invalid — encodes whether the invariant is corroborated by a live ranking signal or is documented-only guidance; `documented-only` + a `firm` model finding is downgraded to `tentative` by the analyzer)
+- `- **Keywords:** comma, separated, keywords` (soft field, defaults to `[]`)
+
+A block missing `Title`, `PrimarySourceUrl`, or a valid `PolicyClass` is dropped by the
+parser — never a fatal error. A `never-encode` block is dropped even when every other
+field is present: it documents the automation boundary for human readers but must
+never reach an analyzer prompt.
+
+## Signatures
+
+### youtube-mentions-ai-visibility
+- **Title:** YouTube mentions are the strongest AI-visibility correlate
+- **Workflows:** ai-visibility
+- **Tactic:** Prioritise earning brand presence and mentions on YouTube (owned channel content, creator partnerships, video PR) as a first-class AI-visibility lever, not an afterthought to written-content SEO.
+- **Invariant:** Across ChatGPT, Google AI Mode, and AI Overviews, YouTube mentions show the single highest correlation (~0.737 Spearman) with brand visibility in AI answers, in a 75,000-brand Ahrefs correlation study.
+- **PrimarySourceUrl:** https://ahrefs.com/blog/ai-brand-visibility-correlations/
+- **PolicyClass:** human-approve
+- **EvidenceGrade:** verified
+- **Keywords:** youtube, ai-visibility, brand-mention, geo, correlation
+
+### branded-mentions-ai-visibility
+- **Title:** Branded web mentions (linked or unlinked) are the strongest non-video AI-visibility signal
+- **Workflows:** ai-visibility
+- **Tactic:** Pursue digital-PR and organic brand-mention placements across the web; unlinked mentions still count toward AI-visibility correlation, so do not discount coverage that omits a hyperlink.
+- **Invariant:** Branded web mentions (linked or unlinked) correlate at 0.66-0.71 with brand visibility in AI answers — the strongest non-YouTube signal in the same 75,000-brand study.
+- **PrimarySourceUrl:** https://ahrefs.com/blog/ai-brand-visibility-correlations/
+- **PolicyClass:** human-approve
+- **EvidenceGrade:** verified
+- **Keywords:** brand-mention, digital-pr, ai-visibility, geo, unlinked-mention
+
+### link-volume-weak-for-ai
+- **Title:** Classic backlink volume is a weak lever for AI visibility
+- **Workflows:** ai-visibility
+- **Tactic:** Do not spend budget on bulk backlink-volume link building as a primary AI-visibility lever; redirect it toward branded mentions and YouTube presence, which correlate far more strongly.
+- **Invariant:** Classic link metrics (backlink count, URL rating) show very weak correlation (~0.218) with AI-answer brand mentions, in the same 75,000-brand study.
+- **PrimarySourceUrl:** https://ahrefs.com/blog/ai-brand-visibility-correlations/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** backlinks, link-building, ai-visibility, weak-correlation
+
+### aio-commercial-intent-shift
+- **Title:** AI Overviews triggers have shifted toward commercial/transactional intent
+- **Workflows:** ai-visibility, rank-track
+- **Tactic:** Prioritise AIO-visibility work on commercial and transactional queries, not only informational content — the trigger mix has moved decisively toward money queries.
+- **Invariant:** AIO intent mix shifted Jan-Oct 2025: informational share fell 91.3%->57.1% of triggers while commercial rose to 18.57% and transactional to 13.94%, per a Semrush 10M+ keyword panel.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/semrush-ai-overviews-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** aio, commercial-intent, transactional, rank-track
+
+### aio-prevalence-volatility
+- **Title:** AI Overview prevalence is highly volatile — plan for churn, not a stable surface
+- **Workflows:** ai-visibility
+- **Tactic:** Re-measure AIO prevalence for target queries on a recurring cadence rather than assuming a one-time audit is durable; build monitoring, not a static report.
+- **Invariant:** AIO prevalence fluctuated hard through 2025: 6.49% of queries (Jan) -> 24.61% (Jul) -> 15.69% (Nov), per the same Semrush 10M+ keyword panel.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/semrush-ai-overviews-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** aio, volatility, monitoring, rank-track
+
+### zero-click-not-collapsing
+- **Title:** Zero-click rate did not spike after AI Overviews — do not over-index on click-loss panic
+- **Workflows:** ai-visibility
+- **Tactic:** Do not justify SEO/GEO investment primarily on a claimed AIO-driven zero-click collapse; the measured effect is small, so frame AI-visibility work around brand exposure and citation share instead.
+- **Invariant:** Zero-click rate slightly DECREASED (33.75%->31.53%) on keywords after AIOs appeared, in a 200K+ keyword before/after study validated against Datos clickstream data.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/semrush-ai-overviews-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** zero-click, aio, clickstream, datos
+
+### aio-citations-decouple-top10
+- **Title:** AIO citations increasingly decouple from organic top-10 rankings
+- **Workflows:** ai-visibility, technical-audit
+- **Tactic:** Do not assume organic top-10 ranking is required for AIO citation; audit AIO citation coverage separately from rank-tracking, since a majority of cited pages now rank outside the top 10.
+- **Invariant:** Only 38% of AIO-cited pages ranked in the organic top-10 as of Jan 2026 (down from ~76% in Jul 2025); 31.2% of cited pages ranked 11-100 and 31.0% ranked beyond top-100, per an Ahrefs analysis of 863K SERPs / 4M AIO URLs.
+- **PrimarySourceUrl:** https://ahrefs.com/blog/ai-overview-citations-top-10/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** aio, citations, top10, technical-audit
+
+### ghost-citations
+- **Title:** Most AI citations are "ghost citations" — cited as a source but never named
+- **Workflows:** ai-visibility
+- **Tactic:** Track citation rate and brand-mention rate as two distinct AI-visibility metrics; a page being cited as a source link does not mean the brand is actually named in the answer.
+- **Invariant:** 61.7% of AI citations are "ghost citations" — the source page is linked but the brand is never named in the answer, across 3,981 domain appearances / 115 prompts / 4 platforms / 14 countries (Semrush/Kevin Indig study).
+- **PrimarySourceUrl:** https://www.semrush.com/blog/the-ghost-citations-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** ghost-citation, citation-rate, mention-rate, ai-visibility
+
+### query-fan-out-coverage
+- **Title:** AI Overviews retrieve via query fan-out — optimize for the fan-out set, not just the head keyword
+- **Workflows:** topical-map, ai-visibility
+- **Tactic:** Build topical-map content that comprehensively covers the fan-out query set around a topic (related sub-questions, comparisons, specifications) rather than optimizing a single page for one head keyword.
+- **Invariant:** AI Overviews retrieve results via query fan-out (generating and retrieving against a broadened set of related sub-queries), so single-keyword-optimized pages under-cover what AIO actually retrieves against.
+- **PrimarySourceUrl:** https://ahrefs.com/blog/ai-overview-citations-top-10/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** fan-out, topical-map, query-expansion, aio
+
+### perplexity-community-weighted
+- **Title:** Perplexity is the most community-weighted AI platform — Reddit dominates its citation mix
+- **Workflows:** ai-visibility
+- **Tactic:** For Perplexity-specific AI-visibility work, prioritise genuine community presence and discussion (Reddit and similar forums) over traditional publisher-style content placement.
+- **Invariant:** Perplexity is the most community-weighted platform in a 680M-citation study: Reddit accounts for 6.6% of ALL Perplexity citations and 46.7% of its top-source citations.
+- **PrimarySourceUrl:** https://www.tryprofound.com/blog/ai-platform-citation-patterns
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** perplexity, reddit, community, platform-specific
+
+### sitefocus-topical-authority
+- **Title:** Topical authority is mechanically scored — site focus and page-topic deviation affect ranking potential
+- **Workflows:** topical-map
+- **Tactic:** Keep new content aligned to the site's established topical focus; treat high-deviation off-topic pages as ranking-risk outliers rather than easy incremental traffic.
+- **Invariant:** The leaked Google Content Warehouse API documents a siteFocusScore (site dedication to one topic) and siteRadius (page deviation from the site's central theme) that operationalize topical authority; high-deviation pages are documented ranking outliers.
+- **PrimarySourceUrl:** https://sparktoro.com/blog/an-anonymous-source-shared-thousands-of-leaked-google-search-api-documents-with-me-everyone-in-seo-should-see-them/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **LiveWeightStatus:** documented-only
+- **Keywords:** topical-authority, sitefocus, site-radius, leak
+
+### schema-entity-aio-lift
+- **Title:** Entity/schema markup measurably lifts AI Overview visibility
+- **Workflows:** schema-audit, ai-visibility
+- **Tactic:** Implement and validate entity-linking schema markup (Organization, Product, FAQ, and related structured data) as a standard part of every technical/schema audit, not an optional add-on.
+- **Invariant:** A documented Schema App case study measured +19.72% AI Overview visibility from entity-linking / schema-markup work — one of the few quantified schema-to-GEO results.
+- **PrimarySourceUrl:** https://www.schemaapp.com/schema-markup/case-study-entity-linking-increases-aio-visibility/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** schema, entity, structured-data, aio-lift
+
+### parasite-seo-placement
+- **Title:** Parasite SEO placement on high-authority third-party hosts
+- **Workflows:** parasite-watch
+- **Tactic:** NEVER encode as an actionable tactic — placing brand content on a high-authority third-party domain (forum, subreddit, unrelated news site) specifically to borrow its ranking signal is a Google site-reputation-abuse policy violation. The Nov 2024 policy update closed the first-party-involvement loophole: no amount of oversight, white-labeling, or licensing exempts this.
+- **Invariant:** This tactic must never be recommended, even as a human-approval-gated option — it is a named policy violation, not a judgment call the suite should surface as an option.
+- **PrimarySourceUrl:** https://developers.google.com/search/blog/2024/03/core-update-spam-policies
+- **PolicyClass:** never-encode
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** parasite-seo, site-reputation-abuse, never-encode, policy-violation

package/skills/bober.seo-internal-linking/SKILL.md

@@ -0,0 +1,87 @@
+---
+name: bober.seo-internal-linking
+description: "Per-workflow SEO signature library for the internal-linking workflow. A data file of discrete, cited SEO tactics read by SeoPlaybookParser (src/seo/parser.ts), not a workflow skill. Encodes leak-grounded siteFocusScore/siteRadius topical-consolidation signals and fresh/top-tier-indexed link value for the internal-linking workflow."
+---
+
+# bober.seo-internal-linking — Internal Linking Signature Library
+
+This skill is a **signature-library** file for the `internal-linking` workflow, not a
+workflow skill in its own right. It is read (as raw markdown text) by
+`SeoPlaybookParser.parse()` (`src/seo/parser.ts`) and turned into typed
+`SeoSignature[]` records consumed by the SEO agent team's retriever
+(`SeoPlaybookRetriever`). It follows the identical block format defined and
+documented in `skills/bober.seo-generic/SKILL.md` ("Signature Block Format") --
+the two files are one executable spec; a block missing `Title` or
+`PrimarySourceUrl`, or tagged `PolicyClass: never-encode`, is DROPPED by the
+parser at load time -- see `.bober/research/research-20260715-ultimate-seo-agents-skills-research.md`
+for the underlying evidence.
+
+The signatures below encode the leaked Google Content Warehouse API's
+`siteFocusScore` / `siteRadius` topical-consolidation attributes, the
+tactical implication that fresh/top-tier-indexed pages pass more link
+value, and the anchor-text demotion signal. Most signatures here are
+**read-only analysis** (auto-safe); the one signature that would execute a
+redirect or page-merge is `human-approve`, since it is a structural site
+change, not an audit -- consistent with research §6's automation boundary.
+
+## Signatures
+
+### sitefocus-internal-consolidation
+- **Title:** Concentrate internal links within topical clusters to reinforce siteFocusScore
+- **Workflows:** internal-linking
+- **Tactic:** Link new and existing pages primarily to other pages within the same topical cluster; avoid diluting internal link equity across unrelated topics on the site.
+- **Invariant:** The leaked Google Content Warehouse API documents a siteFocusScore that rewards a site's dedication to one topic; concentrating internal links within topical clusters reinforces the site-wide signal the leak shows Google is mechanically scoring.
+- **PrimarySourceUrl:** https://sparktoro.com/blog/an-anonymous-source-shared-thousands-of-leaked-google-search-api-documents-with-me-everyone-in-seo-should-see-them/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** sitefocus, topical-cluster, internal-linking, leak
+
+### siteradius-deviation-outlier
+- **Title:** Limit internal link equity flowing to high-siteRadius off-topic outlier pages
+- **Workflows:** internal-linking
+- **Tactic:** When auditing internal-link structure, identify pages that deviate sharply from the site's central topic and deliberately limit the internal-link equity routed to them rather than treating them as easy incremental-traffic targets.
+- **Invariant:** The leaked Google Content Warehouse API documents siteRadius, which scores each page's deviation from the site's central theme; the leak identifies high-deviation pages as documented ranking outliers, so an internal-linking strategy should isolate rather than reinforce them.
+- **PrimarySourceUrl:** https://sparktoro.com/blog/an-anonymous-source-shared-thousands-of-leaked-google-search-api-documents-with-me-everyone-in-seo-should-see-them/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** siteradius, deviation, ranking-outlier, internal-linking
+
+### fresh-toptier-link-value
+- **Title:** Route internal links from freshly-updated, well-indexed hub pages
+- **Workflows:** internal-linking
+- **Tactic:** Prioritise placing internal links to priority pages from hub/pillar pages that are frequently updated and confirmed well-indexed, rather than from stale or thinly-indexed pages.
+- **Invariant:** Mike King's technical analysis of the leak states the tactical implication that links from fresh and top-tier-indexed pages weigh more in Google's ranking systems than links from stale or poorly-indexed pages.
+- **PrimarySourceUrl:** https://ipullrank.com/google-algo-leak
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** fresh-link, top-tier-indexed, link-value, hub-page
+
+### anchor-mismatch-demotion
+- **Title:** Keep internal anchor text on-topic to avoid anchorMismatchDemotion
+- **Workflows:** internal-linking
+- **Tactic:** Audit internal anchor text for relevance to the linked page's actual topic; avoid over-optimized, keyword-stuffed, or topically mismatched anchor text even when it targets a high-value keyword.
+- **Invariant:** The leak documents anchorMismatchDemotion together with IsAnchorBayesSpam, a Penguin-era anchor-text spam classifier; anchor text that does not match the linked page's topic is a documented demotion trigger.
+- **PrimarySourceUrl:** https://ipullrank.com/google-algo-leak
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** anchor-text, anchor-mismatch-demotion, penguin, internal-linking
+
+### authority-consolidation-merge
+- **Title:** Consolidating overlapping pages concentrates authority -- but requires a human-approved redirect/migration
+- **Workflows:** internal-linking
+- **Tactic:** When internal-link analysis surfaces overlapping or cannibalising pages, propose consolidating them into the stronger URL and routing internal links accordingly; do not execute the redirect or migration without a human sign-off, since it is a structural site change.
+- **Invariant:** The leak's siteAuthority and homepagePagerankNs attributes give a mechanical basis for authority-consolidation strategy (merging overlapping pages into the stronger URL to concentrate site-wide authority signals); because consolidation requires redirects and site migration, a human must own the execution decision.
+- **PrimarySourceUrl:** https://ipullrank.com/google-algo-leak
+- **PolicyClass:** human-approve
+- **EvidenceGrade:** single-source
+- **Keywords:** authority-consolidation, redirect, site-migration, human-approve
+
+### topical-cluster-hub-spoke
+- **Title:** Build hub-and-spoke internal linking to concentrate siteFocusScore-scored topical authority
+- **Workflows:** internal-linking
+- **Tactic:** Structure internal linking as hub pages fanning out to member/spoke pages within a topic, rather than flat, undifferentiated cross-linking across the whole site.
+- **Invariant:** The leak's siteFocusScore mechanically rewards a site's dedication to one topic; a hub-and-spoke internal-link structure concentrates topical signal into a small set of pillar pages instead of diffusing it flatly across the site.
+- **PrimarySourceUrl:** https://sparktoro.com/blog/an-anonymous-source-shared-thousands-of-leaked-google-search-api-documents-with-me-everyone-in-seo-should-see-them/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** hub-spoke, topical-cluster, sitefocusscore, internal-linking

package/skills/bober.seo-parasite-watch/SKILL.md

@@ -0,0 +1,121 @@
+---
+name: bober.seo-parasite-watch
+description: "Per-workflow SEO signature library for the parasite-watch workflow. A data file of discrete, cited SEO tactics read by SeoPlaybookParser (src/seo/parser.ts), not a workflow skill. Encodes READ-ONLY detection/monitoring guidance for competitors' parasite SEO on high-DR third-party hosts, plus the Nov-2024 site-reputation-abuse policy, framed exclusively as defender-side awareness. Every policy-violating tactic itself (placement, expired-domain plays, paid links) is encoded ONLY as a never-encode block the parser drops -- see the safety note below."
+---
+
+# bober.seo-parasite-watch — Parasite-SEO Detection Signature Library
+
+This skill is a **signature-library** file for the `parasite-watch` workflow, not
+a workflow skill in its own right. It is read (as raw markdown text) by
+`SeoPlaybookParser.parse()` (`src/seo/parser.ts`) and turned into typed
+`SeoSignature[]` records consumed by the SEO agent team's retriever
+(`SeoPlaybookRetriever`). It follows the identical block format defined and
+documented in `skills/bober.seo-generic/SKILL.md` ("Signature Block Format") --
+the two files are one executable spec; a block missing `Title` or
+`PrimarySourceUrl`, or tagged `PolicyClass: never-encode`, is DROPPED by the
+parser at load time.
+
+**Safety framing (read before editing this file):** parasite SEO is an
+adversarial tactic under a NAMED Google spam policy. Every surviving block
+below is read-only competitor monitoring -- phrased as Detect / Monitor / Flag
+/ Audit, never as an instruction to place content, buy links, or register a
+domain. Every policy-violating tactic itself (placement on a third-party host,
+expired-domain authority plays, paid links) is encoded as a `PolicyClass:
+never-encode` block, which the parser hard-drops at `src/seo/parser.ts:111` so
+it never reaches an analyzer prompt -- it exists here only to document the
+automation boundary for human readers, mirroring the precedent block
+`parasite-seo-placement` in `skills/bober.seo-generic/SKILL.md:163-171`. See
+`.bober/research/research-20260715-ultimate-seo-agents-skills-research.md`
+§3/§6 for the underlying policy evidence.
+
+## Signatures
+
+### parasite-competitor-detection-highdr-hosts
+- **Title:** Detect competitors' parasite SEO on high-DR third-party hosts
+- **Workflows:** parasite-watch
+- **Tactic:** Monitor target SERPs for competitor content placed on high-authority third-party hosts (Forbes, Reddit, Medium, major news outlets, .edu domains) -- keyword-stuffed "best <X>" listicles or review pages outranking dedicated competitor sites are the parasite-SEO signal. Flag newly-appearing parasite entrants for the human operator to review. This is READ-ONLY SERP monitoring, never a placement action.
+- **Invariant:** Parasite SEO (site-reputation abuse) remains widespread across gambling/affiliate SERPs despite the Nov-2024 policy update; continuous SERP monitoring detects new parasite entrants a defender must respond to.
+- **PrimarySourceUrl:** https://developers.google.com/search/blog/2024/11/site-reputation-abuse
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** parasite-seo, detection, site-reputation-abuse, serp-monitoring, competitor
+
+### site-reputation-abuse-policy-awareness
+- **Title:** Maintain policy awareness of the Nov-2024 site-reputation-abuse loophole closure
+- **Workflows:** parasite-watch
+- **Tactic:** Keep the parasite-watch audit report's policy-reference section current with Google's site-reputation-abuse definition; use it as the classification rubric when flagging detected competitor placements, so findings are labeled against the actual named policy rather than an informal "seems spammy" judgment.
+- **Invariant:** Google's Nov 2024 policy update closed the first-party-involvement loophole for site-reputation abuse: no amount of third-party oversight, white-labeling, or licensing arrangement exempts content hosted to exploit another site's ranking signal.
+- **PrimarySourceUrl:** https://developers.google.com/search/blog/2024/11/site-reputation-abuse
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** site-reputation-abuse, policy-awareness, nov-2024, classification
+
+### standalone-section-algorithm-watch
+- **Title:** Watch for Google scoring a site section as "standalone" from the main site
+- **Workflows:** parasite-watch, technical-audit
+- **Tactic:** Audit site sections (subdomains, subdirectories with distinct editorial control or topic focus) for the risk of being algorithmically scored as standalone from the main site's authority; report any section showing "starkly different" content/purpose from the core site as a silent-traffic-loss risk finding, not a policy violation -- this is a ranking-mechanics watch item, not a spam classification.
+- **Invariant:** Google's spam-policy documentation notes that a site section starkly different in purpose from the rest of the site may be scored as if it were a standalone entity, which can silently reduce the traffic that section inherits from the main site's authority -- a distinct risk from an outright policy violation.
+- **PrimarySourceUrl:** https://developers.google.com/search/docs/essentials/spam-policies
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** standalone-section, algorithm-watch, silent-traffic-loss, technical-audit
+
+### parasite-serp-monitoring-new-entrants
+- **Title:** Run continuous SERP monitoring for new parasite-SEO entrants in adversarial verticals
+- **Workflows:** parasite-watch
+- **Tactic:** Schedule recurring (not one-time) SERP checks for target keywords in adversarial verticals (iGaming, affiliate, crypto) to catch new parasite placements as they appear; a single audit snapshot understates the problem because parasite entrants rotate hosts and rankings volatilely.
+- **Invariant:** Industry reporting on iGaming affiliate SERPs documents parasite SEO as an ongoing, shifting problem -- new placements appear and enforcement against them is inconsistent -- making continuous monitoring, not a one-time audit, the correct defensive cadence.
+- **PrimarySourceUrl:** https://www.affiversemedia.com/parasite-seo-is-thriving-in-igaming-and-its-costing-affiliates/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** serp-monitoring, new-entrants, igaming, continuous-audit
+
+### remediation-trap-awareness
+- **Title:** Flag remediation traps before recommending a manual-action response
+- **Workflows:** parasite-watch
+- **Tactic:** When a site under audit has a manual action or reputation-abuse penalty, route any proposed remediation (noindexing the offending section, moving content to a different subdirectory) through human approval; do not auto-recommend noindex as a fix, and explicitly flag that relocating penalized content to a new path may itself be treated as circumvention rather than remediation.
+- **Invariant:** Noindexing a section does not automatically remove an existing manual action, and moving previously-penalized content into a new subdirectory can be treated by Google as an attempt to circumvent the action rather than a genuine fix -- both are documented remediation traps in Google's spam-policy guidance.
+- **PrimarySourceUrl:** https://developers.google.com/search/docs/essentials/spam-policies
+- **PolicyClass:** human-approve
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** remediation-trap, manual-action, circumvention, human-approve
+
+### igaming-parasite-defensive-playbook
+- **Title:** Defend against iGaming parasite-SEO pressure with long-tail targeting and diversified traffic
+- **Workflows:** parasite-watch
+- **Tactic:** For iGaming/affiliate clients facing parasite-SEO pressure on head terms, recommend (human-approval-gated, since it is a strategic content/PR investment) a defensive mix of long-tail keyword targeting the parasite entrants under-cover, genuine E-E-A-T-building PR (real expert bylines, verifiable credentials), and traffic-source diversification so no single SERP category dominates revenue exposure.
+- **Invariant:** Reporting on the iGaming affiliate-marketing parasite-SEO landscape documents affiliates experiencing organic-traffic declines from parasite competitors and recommends long-tail targeting, genuine authority-building, and traffic diversification as the practical defensive response, alongside continuous monitoring.
+- **PrimarySourceUrl:** https://www.businessofigaming.com/parasite-seo-affiliate-marketing/
+- **PolicyClass:** human-approve
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** igaming, defensive-playbook, long-tail, eeat, traffic-diversification
+
+### parasite-seo-placement-boundary
+- **Title:** Parasite-SEO placement on high-DR third-party hosts -- NEVER encode as a tactic
+- **Workflows:** parasite-watch
+- **Tactic:** NEVER encode as actionable. Placing brand or client content on a high-authority third-party domain (forum, subreddit, unrelated news site) specifically to borrow its ranking signal is a Google site-reputation-abuse policy violation, regardless of disclosure or licensing arrangement. Documented here only to mark the automation boundary; the parser DROPS this block so it can never reach an analyzer prompt.
+- **Invariant:** This tactic must never be recommended, even as a human-approval-gated option -- it is a named policy violation, not a judgment call the suite should surface as an option. See the identical precedent in the generic skill's `parasite-seo-placement` block.
+- **PrimarySourceUrl:** https://developers.google.com/search/blog/2024/11/site-reputation-abuse
+- **PolicyClass:** never-encode
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** parasite-seo, never-encode, policy-violation, site-reputation-abuse
+
+### expired-domain-authority-play-boundary
+- **Title:** Expired-domain authority acquisition -- NEVER encode as a tactic
+- **Workflows:** parasite-watch
+- **Tactic:** NEVER encode as actionable. Buying/re-registering an expired domain to inherit its historic authority and backlinks (then redirecting or rebuilding on it) is a NAMED Google spam violation ("expired domain abuse", March 2024). Documented here only to mark the automation boundary; the parser DROPS this block so it can never reach an analyzer prompt.
+- **Invariant:** Expired-domain abuse is a named spam policy -- must never be surfaced as an option, even human-approval-gated.
+- **PrimarySourceUrl:** https://developers.google.com/search/blog/2024/03/core-update-spam-policies
+- **PolicyClass:** never-encode
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** expired-domain, authority-play, never-encode, policy-violation
+
+### paid-links-boundary
+- **Title:** Buying or selling links that pass ranking signal -- NEVER encode as a tactic
+- **Workflows:** parasite-watch
+- **Tactic:** NEVER encode as actionable. Buying links, selling links that pass PageRank-style signal, or exchanging goods/services (including paid posts or sponsorships) specifically for links intended to manipulate ranking is a NAMED Google link-spam violation. Documented here only to mark the automation boundary; the parser DROPS this block so it can never reach an analyzer prompt.
+- **Invariant:** Paid/link-spam schemes are a named spam policy category in Google's living spam-policy reference -- must never be surfaced as an option, even human-approval-gated.
+- **PrimarySourceUrl:** https://developers.google.com/search/docs/essentials/spam-policies
+- **PolicyClass:** never-encode
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** paid-links, link-spam, never-encode, policy-violation

package/skills/bober.seo-rank-track/SKILL.md

@@ -0,0 +1,98 @@
+---
+name: bober.seo-rank-track
+description: "Per-workflow SEO signature library for the rank-track workflow. A data file of discrete, cited SEO tactics read by SeoPlaybookParser (src/seo/parser.ts), not a workflow skill. Encodes AI Overview citation-vs-rank decoupling, AIO prevalence volatility, and the zero-click-rate reality for the rank-track workflow."
+---
+
+# bober.seo-rank-track — Rank Tracking Signature Library
+
+This skill is a **signature-library** file for the `rank-track` workflow, not a
+workflow skill in its own right. It is read (as raw markdown text) by
+`SeoPlaybookParser.parse()` (`src/seo/parser.ts`) and turned into typed
+`SeoSignature[]` records consumed by the SEO agent team's retriever
+(`SeoPlaybookRetriever`). It follows the identical block format defined and
+documented in `skills/bober.seo-generic/SKILL.md` ("Signature Block Format") --
+the two files are one executable spec; a block missing `Title` or
+`PrimarySourceUrl`, or tagged `PolicyClass: never-encode`, is DROPPED by the
+parser at load time -- see
+`.bober/research/research-20260715-ultimate-seo-agents-skills-research.md`
+for the underlying evidence.
+
+The signatures below encode Ahrefs' and Semrush's 2025-2026 rank-tracking
+studies: AI Overview citation increasingly decouples from organic top-10
+position (only 38% of AIO-cited pages rank top-10), AIO prevalence and
+intent mix are highly volatile, zero-click rate has not spiked as claimed,
+and most AI citations are "ghost citations" (linked but unnamed). Every
+signature here is pure read-only monitoring/analysis, so all are
+**auto-safe** -- consistent with research §6's automation boundary.
+
+## Signatures
+
+### aio-citation-rank-decouple
+- **Title:** Track AIO-citation coverage as a metric separate from organic rank -- only 38% of cited pages rank top-10
+- **Workflows:** rank-track
+- **Tactic:** Record AIO-citation status as an independent tracked metric alongside position, not as a proxy derived from rank; a page can be a heavily-cited AIO source while ranking well outside the organic top-10, so rank-only tracking misses most of the citation surface.
+- **Invariant:** An Ahrefs analysis of 863K SERPs / 4M AIO URLs found only 38% of AIO-cited pages ranked in the organic top-10 as of Jan 2026 (down from ~76% in Jul 2025), with 31.2% ranking 11-100 and 31.0% ranking beyond top-100.
+- **PrimarySourceUrl:** https://ahrefs.com/blog/ai-overview-citations-top-10/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** aio-citation, rank-decouple, top10, ahrefs
+
+### zero-click-reality-check
+- **Title:** Zero-click rate slightly decreased after AIOs appeared -- don't over-index on click-loss panic
+- **Workflows:** rank-track
+- **Tactic:** When interpreting rank-tracking data for click-loss risk, do not default to a narrative of AIO-driven zero-click collapse; report the measured, small directional change instead of an assumed large one.
+- **Invariant:** A 200K+ keyword before/after study validated against Datos clickstream data found the zero-click rate slightly DECREASED (33.75% to 31.53%) on keywords after AI Overviews appeared.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/semrush-ai-overviews-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** zero-click, reality-check, datos, clickstream
+
+### aio-prevalence-volatility-cadence
+- **Title:** AI Overview prevalence swings hard -- re-measure on a recurring cadence, not a one-time snapshot
+- **Workflows:** rank-track
+- **Tactic:** Schedule AIO-prevalence re-measurement for tracked queries on a recurring cadence; treat AIO as a churning surface that needs standing monitoring, not a one-time audit finding that stays valid indefinitely.
+- **Invariant:** A Semrush 10M+ keyword panel documents AIO prevalence fluctuating through 2025 from 6.49% of queries (Jan) to 24.61% (Jul) to 15.69% (Nov) -- a large enough swing that a single-point measurement is stale within weeks.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/semrush-ai-overviews-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** aio-prevalence, volatility, monitoring-cadence, semrush
+
+### commercial-intent-trigger-shift
+- **Title:** Track commercial/transactional AIO triggers -- intent mix has moved decisively to money queries
+- **Workflows:** rank-track
+- **Tactic:** Segment rank-tracking reports by intent (informational vs. commercial vs. transactional) and surface the commercial/transactional AIO-trigger trend as a distinct tracked line, since it reflects where the AIO surface is actually shifting.
+- **Invariant:** The same Semrush 10M+ keyword panel documents AIO intent mix shifting Jan-Oct 2025: informational share fell from 91.3% to 57.1% of triggers while commercial rose to 18.57% and transactional to 13.94%.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/semrush-ai-overviews-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** commercial-intent, transactional, aio-trigger-shift, rank-track
+
+### rank-track-monitoring-cadence
+- **Title:** Track position on a standing recurring cadence rather than ad-hoc spot checks
+- **Workflows:** rank-track
+- **Tactic:** Run position tracking as a recurring, read-only monitoring pass on a fixed cadence for tracked queries/URLs, rather than triggering checks only ad-hoc when a stakeholder asks; a recurring cadence is what makes AIO-citation-decouple and volatility findings comparable run-over-run.
+- **Invariant:** Both the AIO citation-decouple finding (38% top-10 alignment, down from ~76% seven months earlier) and the AIO prevalence-volatility finding (6.49%->24.61%->15.69% across 2025) show the ranking/citation surface moving materially within months, which only a recurring monitoring cadence can capture.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/semrush-ai-overviews-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** rank-track, monitoring-cadence, recurring, position-tracking
+
+### ghost-citation-two-metric-tracking
+- **Title:** Track citation-rate and mention-rate as two separate metrics -- most AI citations are "ghost citations"
+- **Workflows:** rank-track
+- **Tactic:** Report AI citation-rate (linked as a source) and brand-mention-rate (named in the answer text) as two distinct tracked metrics rather than treating a citation as evidence the brand was actually named.
+- **Invariant:** A Semrush/Kevin Indig study of 3,981 domain appearances / 115 prompts / 4 platforms / 14 countries found 61.7% of AI citations are "ghost citations" -- the source page is linked but the brand is never named in the answer.
+- **PrimarySourceUrl:** https://www.semrush.com/blog/the-ghost-citations-study/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** verified
+- **Keywords:** ghost-citation, citation-rate, mention-rate, two-metric
+
+### navboost-click-quality-leading-indicator
+- **Title:** Watch good/bad-click proxies as a leading indicator of upcoming rank movement
+- **Workflows:** rank-track
+- **Tactic:** Include click-quality proxies (bounce-back behavior, dwell/engagement signals) as a tracked leading-indicator line item alongside raw position, since click-history shifts can precede a rank change rather than only follow it.
+- **Invariant:** The leaked Google Content Warehouse API documents NavBoost as a click-based re-ranking system tracking goodClicks, badClicks, lastLongestClicks, and unsquashedClicks over a rolling 13-month window; a shift in this click history is a mechanical input to future re-ranking, making it a leading indicator worth tracking alongside position.
+- **PrimarySourceUrl:** https://ipullrank.com/google-algo-leak
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** navboost, click-quality, leading-indicator, rank-track

package/skills/bober.seo-schema-audit/SKILL.md

@@ -0,0 +1,97 @@
+---
+name: bober.seo-schema-audit
+description: "Per-workflow SEO signature library for the schema-audit workflow. A data file of discrete, cited SEO tactics read by SeoPlaybookParser (src/seo/parser.ts), not a workflow skill. Encodes the entity-linking to AI Overview visibility lift (Schema App case study), Google's structured-data policy guidance, and leak-grounded date-consistency signals for the schema-audit workflow."
+---
+
+# bober.seo-schema-audit — Schema Audit Signature Library
+
+This skill is a **signature-library** file for the `schema-audit` workflow, not a
+workflow skill in its own right. It is read (as raw markdown text) by
+`SeoPlaybookParser.parse()` (`src/seo/parser.ts`) and turned into typed
+`SeoSignature[]` records consumed by the SEO agent team's retriever
+(`SeoPlaybookRetriever`). It follows the identical block format defined and
+documented in `skills/bober.seo-generic/SKILL.md` ("Signature Block Format") --
+the two files are one executable spec; a block missing `Title` or
+`PrimarySourceUrl`, or tagged `PolicyClass: never-encode`, is DROPPED by the
+parser at load time -- see `.bober/research/research-20260715-ultimate-seo-agents-skills-research.md`
+for the underlying evidence.
+
+The signatures below encode the quantified entity-linking -> AI Overview
+visibility lift (Schema App case study), Google's own structured-data
+eligibility and policy documentation, and the leak's date-consistency
+guidance applied to structured-data audits. Read-only validation and
+audit tactics are `auto-safe`; generating and deploying structured data at
+scale across a site is a structural change with policy exposure and is
+`human-approve` -- consistent with research §6's automation boundary.
+
+## Signatures
+
+### schema-entity-linking-aio
+- **Title:** Entity-linking / schema markup measurably lifts AI Overview visibility (+19.72%)
+- **Workflows:** schema-audit
+- **Tactic:** Prioritise entity-linking and schema-markup work as a standard, budgeted line item in every schema audit, not an optional add-on -- treat it as one of the few quantified levers with a measured AI Overview visibility outcome.
+- **Invariant:** A documented Schema App case study measured a +19.72% increase in AI Overview visibility from entity-linking / schema-markup work -- one of the few quantified schema-to-GEO (generative-engine-optimization) results in the current evidence base.
+- **PrimarySourceUrl:** https://www.schemaapp.com/schema-markup/case-study-entity-linking-increases-aio-visibility/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** entity-linking, schema-markup, aio-lift, schema-audit
+
+### structured-data-validation
+- **Title:** Validate structured data against Google's eligibility guidance as a standard audit step
+- **Workflows:** schema-audit
+- **Tactic:** Run every audited page's structured data through validation against Google's documented eligibility requirements before flagging it as compliant; treat "present but invalid" markup as a distinct finding from "absent" markup.
+- **Invariant:** Google's structured-data documentation defines validation and rich-result eligibility requirements that markup must satisfy; rich-result eligibility depends on valid, complete markup, not merely its presence.
+- **PrimarySourceUrl:** https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** structured-data, validation, rich-results, schema-audit
+
+### article-date-schema-alignment
+- **Title:** Article schema datePublished/dateModified must align with the visible byline
+- **Workflows:** schema-audit
+- **Tactic:** For every audited article page, compare the structured data's datePublished and dateModified against the visible byline date on the page; flag any mismatch as an audit finding.
+- **Invariant:** Google's Article structured-data documentation specifies datePublished and dateModified properties that are expected to reflect the article's actual publish/update history; misalignment with the visible byline is a documented markup-accuracy issue.
+- **PrimarySourceUrl:** https://developers.google.com/search/docs/appearance/structured-data/article
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** article-schema, datepublished, datemodified, byline-alignment
+
+### leak-date-consistency-schema
+- **Title:** Cross-check structured-data dates against on-page and URL date signals (leak-grounded)
+- **Workflows:** schema-audit
+- **Tactic:** As part of every schema audit, cross-check the structured data's date fields against the visible on-page byline date and any date embedded in the URL; report drift across any of the three as a single date-consistency finding.
+- **Invariant:** Mike King's technical analysis of the leak states the tactical implication that bylineDate, syntacticDate, and semanticDate should be kept consistent across structured data, titles, and URLs; a schema audit is a natural enforcement point for this cross-check.
+- **PrimarySourceUrl:** https://ipullrank.com/google-algo-leak
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** date-consistency, structured-data, leak, schema-audit
+
+### mass-schema-generation-approval
+- **Title:** Mass structured-data generation/deployment across a site requires human approval
+- **Workflows:** schema-audit
+- **Tactic:** Do not auto-deploy generated structured data at scale across a site; generate proposed markup, then route it through human review before publication, since bulk deployment is a structural site change with policy exposure.
+- **Invariant:** Google's General Structured Data Guidelines state that structured data must accurately reflect the content of the page it's on and penalise markup that does not match visible content; generating and deploying markup at scale across many pages without per-page review is a policy-exposed structural change, so a human must approve mass generation before deployment.
+- **PrimarySourceUrl:** https://developers.google.com/search/docs/appearance/structured-data/sd-policies
+- **PolicyClass:** human-approve
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** mass-schema-generation, structured-data-policy, human-approve, sd-guidelines
+
+### entity-schema-org-product-faq
+- **Title:** Implement Organization/Product/FAQ entity schema to strengthen the entity graph behind the AIO lift
+- **Workflows:** schema-audit
+- **Tactic:** Where absent, implement Organization, Product, and FAQ structured data (as applicable to the page type) as the concrete entity-linking work behind the measured AI Overview visibility lift, prioritising pages that already rank or are AIO-eligible.
+- **Invariant:** The Schema App case study's +19.72% AI Overview visibility lift was produced by entity-linking / schema-markup work; Organization, Product, and FAQ schema are the standard entity-graph structured-data types that operationalize entity linking on a page.
+- **PrimarySourceUrl:** https://www.schemaapp.com/schema-markup/case-study-entity-linking-increases-aio-visibility/
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** single-source
+- **Keywords:** organization-schema, product-schema, faq-schema, entity-graph
+
+### sd-policy-compliance-audit
+- **Title:** Audit for structured-data policy violations (markup that doesn't reflect visible content)
+- **Workflows:** schema-audit
+- **Tactic:** Check every page's structured data against the visible content it describes and flag any markup that adds information not present on the page, or that is hidden/irrelevant to the user, as a policy-compliance finding -- do not wait for a manual action to surface it.
+- **Invariant:** Google's General Structured Data Guidelines document policy violations including markup that doesn't reflect the page's visible content and irrelevant or misleading markup; a schema audit should proactively check for these violations.
+- **PrimarySourceUrl:** https://developers.google.com/search/docs/appearance/structured-data/sd-policies
+- **PolicyClass:** auto-safe
+- **EvidenceGrade:** primary-unverified
+- **Keywords:** structured-data-policy, compliance-audit, sd-guidelines, schema-audit