One owned URL: AI-search evidence lanes
Keep page diagnostics, Google generative impressions, Bing-supported citations, referral visits and answer sightings separate. Owner sheet first; no property data, score or fifth card.
Keep page diagnostics, Google generative impressions, Bing-supported citations, referral visits and answer sightings separate. Owner sheet first; no property data, score or fifth card.
Approved-article social drafts and landing-page experiment preflight have different local ceilings. Named review and separate connected-action authority; no fifth card or tested vendor benefit.
Both manual routes needed numeric caveat corrections in unblinded self-review. No comparative advantage or fifth card; maintained template first.
The source offers planning prompts and an approximate table, not a power plan or winner. Worksheet and named analyst first; no fifth card.
Social Content supports LinkedIn and X draft formats, but its wider workflow includes local reads and account actions. Template first; no fifth card or tested outcome.
Three mutable selected-file links match their labels. The primary emails link now points to inspected v2.11.37; connector resolution is not a browser redirect or package test.
The separate source can draft a sampled inventory. Estimated traffic, apparent gaps and impact order need source, market, missing-data, paired-coverage and content-owner checks.
Only owner-confirmed, dated site observations can enter a separate content brief. Query rank is not coverage or priority; customer need and capacity need independent evidence.
Five gates separate selected files, chosen-host discovery, effective access, claim review and publication approval. This synthesis adds no new source or host test; template first.
Codex and Claude Code document different local skill routes and permissions. Neither host or the named-skill/ZIP payload was tested. Template first; inspect files and authorize context before install.
Three directory presentations supply source leads, not a shared quality measure, verified package or install clearance. Four conditional cards and two nonrecommendations remain.
Complete 2.1.0→2.1.4 selected-file comparison adds reference routing, local Learnings use, outbound referral and mutable-main tool links; the local draft and ESP stops remain separate.
GitHub lists v2.11.37 emails-source.zip with a reported digest; tagged source displays 2.1.4. ZIP bytes and manifest were inaccessible, so no internal path is verified present or missing.
Mutable selected text displayed 2.3.1. Full 2.1.0 and observed 2.3.1 endpoints compared; the exact full 2.1.0→2.2.0 intermediate delta remains partial. Input-dependent draft or point edit; source claim directions do not replace pool review gates.
Named-skill CLI and ZIP scope are documented, but no payload was inspected. Copywriting selected text displayed 2.2.0; source directs AI-tells check, while this catalogue treats it as optional style only.
Read linked source pages. Cross-skill style reference and separate credential, recipient-data and send guides do not authorize implementation; single-skill inclusion remains unknown.
Four conditional active cards and two historical nonrecommendations retained. Standalone answer now names the removed skills; dated emails evidence separates source examples from our five-email subset. Saved-record audit only.
Retain draft scaffold, owner template first. AI-tells is style review, not detection; hold unsupported claims, proof, rights and CTA links before a separate publish decision.
Five post-signup emails are our subset. Owner template first; timing, exits and metrics are hypotheses, with named proof/CTA review and separate ESP approval. Mutable 2.1.3 observed, not pinned.
Retain a sampled, human-confirmed starting set. Tangential results are normally excluded; query-frequency rank is not threat priority. Versioned owner sheet first; no install advice.
Keep as a conditional draft scaffold, not a validated prioritizer. Use a versioned brief/template first for one content month; no install advice.
Removed from active choice. Its X/38, ratings and fixed priorities do not validate search quality or AI citation. Use an unscored owner checklist for one authorized URL; no install or deployment.
Removed from active choice. Selected source gives stop/scale advice using unsupported profitability and significance rules. Governed worksheet and named analyst first; no account action.
Copywriting and emails have file-attributable 2.1.0 change commits, not verified latest selected-file pins. Emails 2.1.1 pin and full diff remain open; no install or action.
Four-route choice and six-row decision/access/review matrix. The mutable emails file displayed 2.1.1; current pin and inventory unknown. No package test.
Start with a brief/template and reviewed prompt for one local draft. Copywriting 2.1.0 is a conditional candidate; current pin, inventory, numeric assertions, install and quality remain unverified.
Defer a skill search. Dated before/after review for one known rival is coherent, but a diff, sheet, prompt or monitor may suffice. No package or monitor test.
Defer a skill search. Comparing an approved brief with proposed setup is distinct from ongoing UTM governance. Local findings are not live launch clearance; six cards remained at that date.
Defer a package search. Brief need is plausible, but templates, prompts and connected workspaces remain viable; no seventh card or package test.
Two more source files miss the offline brief’s audit, trace and owner-approval gates. Search closed for now; no seventh card or test.
Two original source files inspected. Neither meets the offline briefing audit and approval gates; no seventh card or test.
Search for a bounded offline briefing skill, not a seventh card or install recommendation. No reporting package was inspected or tested.
No substantive claim change evidenced. All six current-day directory listings were unavailable; two file pages and the Superamped README were stale.
Keeps Core Haines emails only as a conditional, draft-only candidate. RampStack email-sequences is excluded; no package or email action was tested.
No selected-file SHA/date or tag finding was established for the four named paths. The stop before an install recommendation remains.
Sets a stop procedure for provenance, data, host assumptions, and human approval. No package or host behavior was tested.
Captures Copywriting selected-file provenance and release evidence; four Superamped selected-file commit/date and tag findings remained unresolved.
Compares the first five records by output, access, data, and review boundary. It adds no source inspection or testing.
Records original sources, stated requirements, license, data surface, review boundaries, and limits for the five selected skills.
| Date | Research | Summary | Evidence status |
|---|---|---|---|
| 31 Oct 2026 Run 39 | One owned URL: AI-search evidence lanes | Use an owner-maintained evidence sheet. Google generative impressions, Bing-supported citations, referral visits and screenshots are different observations. | Official source and method review; no property, URL, export, answer or outcome examined |
| 30 Oct 2026 Run 38 | Two jobs, two stops before connected action | Local social drafts stop before account action; an experiment memo stops before sizing without real inputs or launching. Buffer and Statsig describe possible connected steps, not demonstrated gains. | Saved selected-source reads and official product docs; no account, real brief, skill, vendor workflow or outcome tested |
| 29 Oct 2026 Run 37 | One A/B-test report, two local social-draft routes | Template and manually source-guided route each produced LinkedIn and four X local drafts. Both needed numeric caveat corrections; no comparative fidelity or reviewer-effort advantage was shown. | One published report; unblinded author review, no independent review, skill execution, account action or outcome |
| 28 Oct 2026 Run 36 | Landing-page A/B-test preflight: plan before a skill | Source prompts can scaffold a planning memo; worksheet and named analyst are the first route. Rounded sample table is arithmetically plausible but not an actual power plan. No fifth card. | Mutable selected source and linked references; no immutable pin, approved inputs, installation, live test, result or winner |
| 27 Oct 2026 Run 35 | One approved article → two social drafts | Original Social Content source supports local LinkedIn and X draft shapes. An article-to-post template and reviewed prompt remain the first route; no fifth card. | Mutable cached source, no selected-file pin, full payload, host permission, approved article, account action or outcome test |
| 24 Oct 2026 Run 34 | Four active source links: one tag alignment | Connector resolved four original selected files. The tagged email citation now matches 2.1.4 evidence; three other links remain as labeled. | No browser origin status or redirects, latest path dates, full package, host or outcome test |
| 23 Oct 2026 Run 33 | Optional Competitor Content Analysis: source and evidence limits | Separate optional source, not Content Strategy or Competitor Discovery. Sampled URLs may inform a brief; traffic estimates, gaps and impact order remain hypotheses until checked against dated, paired coverage and customer relevance. | Mutable selected file and related source text; no immutable pin, current full inventory, installation, target-site analysis or outcome test |
| 22 Oct 2026 Run 32 | Competitor leads to content: handoff ceiling | Manual, claim-level transfer of confirmed site observations only. Separate content inventories, customer evidence and capacity before any topic order. | Two mutable selected files read; no immutable pin, complete inventory, two-skill run, customer input or outcome test |
| 21 Oct 2026 Run 31 | Copywriting pre-install evidence ledger | Five distinct stops: exact selected payload, chosen-host discovery, effective permissions, authorized brief and claim review, then separate publication approval. Owner template first; decisions unchanged. | Prior source and host-documentation synthesis only; no new source, package, host, context, draft or outcome test |
| 20 Oct 2026 Run 30 | Copywriting host-documentation boundary | Official Codex CLI/IDE and Claude Code docs describe distinct local folders, invocation, supporting files and permission stops. The maintainer’s command and Claude-web ZIP design are not installed-payload evidence; owner template first. | No CLI or ZIP bytes, host invocation, references, effective policy, local context or output test |
| 19 Oct 2026 Run 29 | Directory signals versus selection evidence | One result per site: a labelled quality score, A–Z listing with safety label, and CLI-install telemetry with audit badges are discovery signals. Original source and smaller authorized route first; active and removed cards unchanged. | No revision pin, full payload, host permission, fit, audit engine or outcome test; one safety detail inaccessible |
| 18 Oct 2026 Run 28 | Tagged emails selected-file delta | Complete net 2.1.0→2.1.4 comparison: reference-before-drafting, Learnings, outbound referral and mutable-main implementation links. Template-first local draft and separate recipient-policy/ESP stop remain. | Selected files complete; exact intermediate delta, archive, install, host and outcomes unverified |
| 17 Oct 2026 Run 27 | Email release-asset boundary | Published source ZIP asset identified with GitHub-reported SHA-256; tagged emails source displays 2.1.4 and names in-skill, cross-skill and implementation paths. Conditional owner-template-first draft and separate recipient-policy/ESP stop remain. | ZIP bytes and manifest inaccessible; no entry verified present or missing, no local digest computed; CLI-selected payload and host behavior unknown |
| 16 Oct 2026 Run 26 | Copywriting source-delta boundary | Mutable source displayed 2.3.1 on 16 October. Compared full 2.1.0 and observed 2.3.1 endpoints; 2.2.0 was a dated 15 October observation. Point edits request changed lines/reasons; pool claim, rights, CTA and publication gates remain distinct. | Exact full 2.1.0→2.2.0 delta partial; latest pin, inventory, payload, host behavior and outcomes unverified |
| 15 Oct 2026 Run 25 | One-skill install documentation boundary | Documented named-skill selection and ZIP design do not verify CLI or archive payload. Four conditional cards and two historical nonrecommendations remain; owner template first, no install or action. | Copywriting mutable selected text 2.2.0; latest pin, full delta, inventory, payload and host behavior unverified |
| 14 Oct 2026 Run 24 | Welcome-email linked-route boundary | Read selected and linked sources; keep five-message Markdown draft conditional, owner template first, and implementation routes stopped. | Mutable 2.1.3 selected text; single-skill inclusion, latest pin, full diff and host/provider behavior unverified |
| 13 Oct 2026 Run 23 | Cross-card decision consistency | Four conditional active cards and two historical nonrecommendations retained. Corrects answer omission and dated emails evidence wording; no new package decision. | Saved-record and local-surface audit; no fresh source, install or route test |
| 12 Oct 2026 Run 22 | Copywriting claim/proof boundary | Owner template first. The source describes sectioned copy but does not verify conversions or AI detection. Hold unsupported claims, testimonials, comparisons, guarantees and CTA links; separate rendered-page review and publication decision. | Mutable selected text 2.1.0; latest pin, inventory, install and output quality unverified |
| 11 Oct 2026 Run 21 | Welcome-email draft boundary | Retain conditional five-email post-signup welcome hypothesis, owner template first. Proof, CTA and recipient rules require named review and later ESP approval. | Mutable selected text 2.1.3 observed; latest pin, full diff and inventory unknown; no draft or send test |
| 10 Oct 2026 Run 20 | Competitor Discovery ranking boundary | Retain a dated, sampled public-web starting set after owner confirmation. Tangential candidates are normally dropped; frequency is a discovery heuristic, not competitive priority. Versioned sheet first. | Mutable selected source and stale directory; no pin, current inventory, install or same-job comparison |
| 9 Oct 2026 Run 19 | Content Strategy priority boundary | Keep draft structure only; X/10 weights, buyer-stage labels and sampled gaps are hypotheses. Versioned brief/template first. | Mutable selected source, stale directory listing; no selected-file pin, install, plan or outcome test |
| 8 Oct 2026 Run 18 | Search Page Audit score boundary | Remove the scored card; use an owner-maintained, unscored checklist for one authorized public URL. | Source-claim audit only; no selected-file pin, install, live-page or outcome test |
| 7 Oct 2026 Run 17 | Ad Campaign Analyzer decision boundary | Selected source gives immediate stop/scale instructions with unsupported profit and significance claims; remove active card, retain dated record. | No installation, export, account or outcome test; SHA/date, tag and inventory unresolved |
| 6 Oct 2026 Run 16 | Core Haines changed-source check | Two selected files have attributable 2.1.0 change commits; Copywriting gains an AI-tells reference, emails adds a cross-skill read and mutable tool links. | Not latest pins; emails 2.1.1 SHA/date, full diff and current inventory open; no test |
| 5 Oct 2026 Run 15 | Choose a route before the six skills | Compare local brief/prompt, host-native, conditional third-party package and governed connected action; six rows retain access and human stops. | Only emails selected file newly viewed: mutable 2.1.1; no current pin, inventory or package test |
| 4 Oct 2026 Run 14 | Approved landing-page copy: skill or simpler route? | One authorized local draft: start with a maintained brief/template and reviewed prompt. Reevaluate Copywriting only against recurring omissions and full package, host, data and review gates. | Current 2.1.0 source viewed; no selected-file pin, inventory, install, quality or outcome test |
| 3 Oct 2026 Run 13 | Competitor messaging-change brief job decision | Defer: a dated source-paired review is coherent, but simpler comparison methods and monitors remain credible; page differences do not prove strategy. | Qualitative job assessment; no package or monitor test |
| 2 Oct 2026 Run 12 | Prelaunch setup QA and UTM governance decision | Separate brief-to-setup comparison from taxonomy governance. Defer a portable-skill search; a spreadsheet and connected QA remain credible. | Qualitative job assessment; no package or workflow test |
| 1 Oct 2026 Run 11 | Two-channel launch-brief job decision | Defer: an offline brief is coherent, but no evidence shows a portable skill improves on a template, reviewed prompt or connected tool. | Qualitative job assessment; no package search or test |
| 30 Sep 2026 Run 10 | Provenance-first reporting search | ClawFu requires dependencies and a script; Digital Marketing Pro has teammate review but not owner release approval or full metric reconciliation. Bounded search closed. | Source-text fit only; no seventh card or test |
| 29 Sep 2026 Run 9 | Monthly reporting skill shortlist | Anthropic accepts pasted metrics but lacks mandatory audit and approval; Skill Me instructs sending. No seventh card. | Source-text fit only; no package or workflow test |
| 28 Sep 2026 Run 8 | Monthly cross-channel briefing job | Defines an offline, human-reviewed aggregate-export brief and tests for a later package search. Catalogue stayed at six at that date. | Qualitative job signals; no package inspection or workflow test |
| 27 Sep 2026 Run 7 | Six-entry source and package-surface recheck | No substantive claim change evidenced; six selected-directory listings not current-day verified. | Cached listing gap; no package or host test |
| 26 Sep 2026 Run 6 | Lifecycle/email candidates under the low-access standard | Core Haines emails is conditional and draft-only; RampStack email-sequences is excluded. | Source-inspected; no package, host, or email-action test |
| 25 Sep 2026 Run 5 | Superamped selected-file provenance recheck | Rechecks four selected paths; no selected-file SHA/date or tag finding was established. | Provenance unresolved; no install recommendation |
| 24 Sep 2026 Run 4 | Safe evaluation boundary and host assumptions | Defines stop conditions for provenance, data, host assumptions, and external action. | Documentation only; no package or host test |
| 23 Sep 2026 Run 3 | Pinned provenance and freshness recheck | Captures Copywriting selected-file provenance and release evidence; four Superamped findings remain unresolved. | Provenance recheck; no package test |
| 22 Sep 2026 Run 2 | Compare five source-inspected marketing skills | Decision comparison using the first five records; no new inspection or test. | Comparison of existing evidence |
| 22 Sep 2026 Run 1 | Source-inspected portable marketing skills | Original-source review of five selected skill packages. | Source-inspected |
Calendar placement is by research date, not publication time.