← OmnekyCONTENT HISTORY

Update to Omneky

Snapshot Oct 27, 2026 · 00:28 UTC · version 2.0.31

Collection source: downloaded plugin package.

WHAT CHANGED

Payment or plan references changed

Full description — Now says: get_products, import_product_from_url, and get_products by name (list my products, import this product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly...

Observed in published text. Live prices and checkout terms have not been verified by this change.

Save Omneky and check future changes in Following.

Up to 3 plugins free · no card required.

All fields and original values.

Full description

Before

Oct 24, 2026 · 13:12 UTC

list_brand_products, identify_product_from_url, and get_product_details by name (list my products, import this product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly performance report, HubSpot, and read-only budget reallocation. Do...

After

Oct 27, 2026 · 00:28 UTC

get_products, import_product_from_url, and get_products by name (list my products, import this product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly performance report, HubSpot, and read-only budget reallocation. Do not use for fir...

Skill instructions

Before

Oct 24, 2026 · 13:12 UTC

list_brand_products, identify_product_from_url, and get_product_details by name (list my products, import this Ahrefs, GSC. The connected Omneky MCP server exposes the names below. Call them **by those exact names** when they appear ...

After

Oct 27, 2026 · 00:28 UTC

get_products, import_product_from_url, and get_products by name (list my products, import this Ahrefs. Use this when the user asks how their ads perform, which ads to make more of, or wants to list or import products. One job: ranked...

Supporting files

Before

Oct 24, 2026 · 13:12 UTC

[]

After

Oct 27, 2026 · 00:28 UTC

[{"relative_path":"references/playbook.md","size_in_bytes":13193}]

Compare saved observations

Download comparison JSON

SKILL.md text changes

− removed · + added

--- before
+++ after
@@ -1,8 +1,8 @@
 ---
 name: omneky-analytics-and-catalogue
 description: >-
-  Call list_brand_products, identify_product_from_url, and
-  get_product_details by name (list my products, import this
+  Call get_products, import_product_from_url, and
+  get_products by name (list my products, import this
   product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly
   performance report, HubSpot, and read-only budget
   reallocation. Do not use for first-run onboarding
@@ -16,75 +16,56 @@
   ads, ROAS drop, CPA drop, CTR, weekly performance report,
   fatigue refresh, budget reallocation, waste pruner, audience
   targeting, target CPA, HubSpot attribution, ABM, Semrush,
-  Ahrefs, GSC.
+  Ahrefs.
 ---
 
 # Ad performance and product catalogue
 
-The connected Omneky MCP server exposes the names below. Call them
-**by those exact names** when they appear on this session's
-`tools/list`. Metrics and reporting reads work without catalogue
-tools. Catalogue names (`list_brand_products`, `get_product_details`,
-`identify_product_from_url`, and the import / create / upsert tools)
-are optional. If they are not listed, say that capability isn't
-available on this connector. Do not invent a catalogue or claim
-Omneky has no product list.
-
-This skill is analysis and catalogue only. Competitor
-positioning → `omneky-competitor-analysis`.
-<!-- host:begin name="ab-tests-routing" -->
-A/B tests → `omneky-ab-tests`.
-<!-- host:end -->
-Weekly plans → `omneky-marketing-plan`.
-<!-- host:begin name="creative-skill-routing" -->
-Image ads → `omneky-image-ads`. Product video →
-`omneky-product-video`. Edit /
-resize → `omneky-edit-and-resize`. Launch →
-`omneky-launch-and-connect`. Credits → `omneky-billing`.
-<!-- host:end -->
+Use this when the user asks how their ads perform, which ads to make
+more of, or wants to list or import products. One job: ranked
+evidence and catalogue records. No render, no spend change.
+
+Call tools by their exact names when they are on `tools/list`.
+Catalogue tools are optional; if they are missing, say that
+capability isn't available on this connector.
 
-Handoff fields:
-[handoff-contract](../_shared/references/handoff-contract.md).
-Sourced ids (how to obtain / what to do if absent):
-[preconditions](../_shared/references/preconditions.md).
+Handoff fields: [handoff-contract](../_shared/references/handoff-contract.md).
+Sourced ids (how to obtain / what to do if absent): [preconditions](../_shared/references/preconditions.md).
 <!-- host:begin name="scan-tools-routing" -->
-Acceptance gates (for a later creative):
-[creative-quality-gates](../_shared/references/creative-quality-gates.md).
-Structured brief (for a later creative):
-[creative-brief](../_shared/references/creative-brief.md).
-Other intents:
-[capability-registry](../_shared/references/capability-registry.md).
-Write for:
-[marketer-persona](../_shared/references/marketer-persona.md).
+Acceptance gates (for a later creative): [creative-quality-gates](../_shared/references/creative-quality-gates.md).
+Structured brief (for a later creative): [creative-brief](../_shared/references/creative-brief.md).
+Other intents: [capability-registry](../_shared/references/capability-registry.md).
+Write for: [marketer-persona](../_shared/references/marketer-persona.md).
 <!-- host:end -->
 
 ## Goal
 
-Ranked evidence, catalogue records, and connector reads. No render. No spend mutation. Keep the weekly performance report.
+Ranked evidence, catalogue records, and connector reads.
 
 ## Prerequisites
 
 | Need | Check | If missing |
 |---|---|---|
 | Brand | `list_brands` | Ask which; do not invent |
-| Coverage | Metric tool for the stated window | Empty is missing data; do not report zeros |
-| Winners / diagnosis | `get_performance_breakdown` / `get_daily_metrics` / `get_performance_movers` | Narrow the window |
-| Product | `list_brand_products` when listed | Import via `identify_product_from_url` → scrape → finalize. If catalogue tools are not on `tools/list`, say that capability isn't available on this connector |
-| HubSpot | `get_hubspot_connection_status` | Skip CRM joins; say disconnected |
-| Semrush / Ahrefs / GSC | matching status tool | Skip that source |
-| Budget apply | (read-only here) | `omneky-launch-and-connect` after a yes (`set_ad_budget` / `set_ad_entity_status`) |
+| Coverage | `get_performance_summary` (`include_coverage=true`) | Say there is no imported data; do not report zeros |
+| Winners / diagnosis | `get_performance_breakdown` / `get_performance_summary` / `get_performance_movers` | Narrow the window |
+| Product | `get_products` when listed | Import via `import_product_from_url` (`url`), then `save_imported_product` (`draft_id`). If catalogue tools are not on `tools/list`, say that capability isn't available on this connector |
+| HubSpot | `list_connections` (`connector="hubspot"`) | Skip CRM joins; say disconnected |
+| Semrush / Ahrefs | matching status tool | Skip that source |
+| Budget apply | (read-only here) | `omneky-launch-and-connect` (`set_ad_budget` / `set_ad_status`; the host confirms) |
 
 ## Flow
 
-Lead with the answer (ROAS, CPA, CTR, spend, lift), one recommendation, and confidence. Recipes: winners, fatigue, weekly performance report, ROAS/CPA drop, waste pruner (evidence), targeting evidence, target-CPA plan, HubSpot attribution / ABM brief, catalogue import. Do not open a separate memo skill.
-
-Collect `source_evidence` from the metric call you just ran: metric,
-window (start and end dates, calendar or rolling), channel, ad or
-creative ids, and sample size when handing a performance-informed
-brief to creative or a plan-driven launch. Collect `approved_claims`
-from the user or the product page / catalogue copy they pointed at;
-each claim needs its source. Do not invent a claim. Plain
-user-directed edits and launches do not need `source_evidence`.
+1. Resolve the brand and confirm coverage for the window.
+2. Run the recipe the user asked for: winners, fatigue, weekly
+   report, ROAS/CPA drop, waste pruner (evidence only), targeting
+   evidence, target-CPA plan, HubSpot attribution, ABM brief, or
+   catalogue import. Details: [playbook](references/playbook.md).
+3. Lead with the answer (ROAS, CPA, CTR, spend, lift), one
+   recommendation, and confidence.
+4. When handing off to creative or a plan-driven launch, attach
+   `source_evidence` (metric, window, channel, ad ids, sample size)
+   and `approved_claims`, each with its source. Never invent a claim.
 
 ## When to use
 
@@ -107,274 +88,30 @@
   `omneky-product-video`)
 <!-- host:end -->
 - Launching or connecting campaigns (`omneky-launch-and-connect`)
-- Plan, credit history, or card on file (`omneky-billing`)
+- Credit balance or plan upgrade (`omneky-billing`)
 <!-- host:begin name="onboarding-agency-when-not" -->
 - First-run setup / first ad today (`omneky-onboarding`)
 - Cross-brand / all clients (`omneky-agency`)
 <!-- host:end -->
 - No product-delete tool
 
-## Winning ads → next batch
-
-Stay on the chosen brand for every call below. Empty rows are missing
-coverage, not a reason to silent-switch brands.
-
-1. Reuse the `brand_id` already chosen in this conversation. Unknown:
-   `list_brands` (pass `search` when the user named the brand). Several
-   matches: ask which one and wait. One clear match: say the name and
-   id, then keep that `brand_id`.
-2. Pick the date window before any metric call.
-   - User named dates: use those inclusive `YYYY-MM-DD` days.
-   - "This week" means the calendar week to date, not the last 7
-     days. Use last 7 days only when the user says "last 7 days".
-     Movers keep `period_days` (default 7) and do not take
-     `date_range`. A winner list with no dates uses the last 30
-     days. Always say which window you used.
-   - This surface has no latest-date tool. Do not invent one.
-3. Coverage: an empty metric result means there is no imported
-   data for that brand, window, or channel.
-   - Do not report zeros. Narrow to a shorter or more recent window,
-     or stop and say there is no imported data.
-   - Empty metrics are missing data, not a zero ROAS, spend, or CTR.
-4. Then, always passing that `brand_id` (it is optional in some
-   schemas; send it anyway so results stay on the chosen brand):
-   - Ranked winners: `get_performance_breakdown` with `dimension`
-     (creative, campaign, ad, or channel), `date_range` as two inclusive
-     dates `["YYYY-MM-DD", "YYYY-MM-DD"]`, `sort_metric` (the metric to
-     rank by, such as `spend` or `roas`; the schema does not list
-     values), `sort_order` (`desc` for winners), and
-     `results_per_page` (10 is enough for a winner list). Pass
-     `filters` as `[]` when not filtering. There is no `metrics`
-     argument; set `include_all_metrics` to `true` when you need every
-     column.
-   - Day-by-day: `get_daily_metrics` with `date_range` (same shape) and
-     `filters` as `[]` when not filtering. There is no `metrics`
-     argument.
-   - What changed vs the prior period: `get_performance_movers` with
-     `dimension`, `metric`, `end_date` (last day of the current window,
-     `YYYY-MM-DD`), and `period_days` (default 7). It builds the prior
-     window. Do not send `date_range` to this tool.
-   - Name the user said: `search_reporting_values` with `dimension` and
-     `q` (the text). It returns labels, not metrics. Put the exact value
-     in a later filter. Operators are `includes` or `excludes` only
-     (`condition`, `dimension`, `value` as a list of strings). Do not
-     invent `=`, `in`, or `gt`.
-   - Current allocated daily budget (not spend, not historical ROAS):
-     `get_channel_budget` with `channels`. Totals plus per-campaign
-     `breakdown` / `basis` when the backend returns them. A disconnected
-     or skipped channel is a per-channel error, not an empty total.
-   - Fill any other required field from the tool's own schema and
-     earlier results in this session. Do not ask the user for internal
-     ids.
-5. Never sum `selected_conversion_metric_value` across channels. That
-   field is the brand's goal and it differs per channel (purchases on
-   one channel, conversions on another). Group by channel, or use a
-   named metric (`purchases`, clicks, spend, ROAS, CTR).
-6. Minimum evidence (also in the handoff contract). State the
-   sample next to every conclusion. Fewer than about 1,000
-   impressions, 30 clicks, or 10 conversions per ad is thin — say
-   "too little data to conclude" and show the numbers. Attribution
-   differs by channel; do not mix attributed totals or sum
-   `selected_conversion_metric_value` across channels. An aggregate
-   ratio is summed numerators over summed denominators, never the
-   mean of per-row ratios.
-7. `get_recommendations` when the user asks what to try next. This is
-   advice, not the metric table from steps 4–5.
-   - Required: `brand_id`.
-   - `start_date` and `end_date` are inclusive `YYYY-MM-DD`. Omit both
-     for the last 30 days. A window longer than 30 days is capped at
-     30. `metric` defaults to `spend`.
-   - `status=timeout`: narrow the dates. Do not retry the identical
-     call in this turn.
-<!-- host:begin name="make-next-batch" -->
-8. To **make** the next batch, hand the winning product or concept to
-   `omneky-image-ads` or `omneky-product-video` with the handoff
-   contract (brand, product, source evidence, approved claims,
-   `source_ad_ids`, and one hypothesis per requested variant). This
-   skill does not generate creatives. Launch is a later handoff. A
-   weekly plan is `omneky-marketing-plan`, not a reconstructed
-   generate/launch sequence here.
-<!-- host:end -->
-
-<!-- host:begin name="performance-to-creative" -->
-## Performance-to-creative flows
-
-Evidence first, then a handoff. Do not invent ROAS. Do not generate
-here. Each requested variant is one hypothesis: source ad, what
-stays constant, what changes, audience / placement, success metric.
-
-- **fatigue-refresh.** `get_performance_movers` on Meta (falling
-  CTR or ROAS; state the window and sample). Pick up to 3 fatigued
-  ads with enough evidence. Hand 3 replacement hypotheses to
-  `omneky-image-ads` (or `omneky-product-video` when they asked for
-  video).
-- **top-roas-variations.** `get_performance_breakdown` sorted by
-  `roas`. Take the verified winners. Hand N distinct hypotheses
-  (N = the count they asked for) to `omneky-image-ads`.
-- **paused-campaign-from-winners.** After the creative skill
-  returns accepted assets, hand `omneky-launch-and-connect` with
-  those URLs. Launch stays `PAUSED` unless they explicitly say to
-  go live.
-- **weekly-performance-report.** Same coverage check, then
-  `get_performance_breakdown` + `get_daily_metrics` +
-  `get_performance_movers`. State the window. Thin samples stay
-  "too little data to conclude." Do not declare an A/B winner
-  here (`omneky-ab-tests`).
-- **budget-reallocation.** Read `get_channel_budget` and the
-  winner / mover tables. Recommend a move (from / to / amount /
-  why). This skill is read-only. Applying `set_ad_budget` is
-  `omneky-launch-and-connect` after one yes. Do not mutate spend
-  here.
-
-Thin samples stay suggestions: say "too little data to conclude"
-and still show the numbers.
-<!-- host:end -->
-
-- **roas-cpa-drop-diagnosis.** Coverage first. Then
-  `get_performance_movers` (falling ROAS or CPA),
-  `get_performance_breakdown`, and `get_daily_metrics`. Explain
-  with evidence only (window, sample, channel). Confidence is
-  thin when samples are thin.
-- **waste-pruner (evidence).** Rank by spend with weak CTR / ROAS
-  on `get_performance_breakdown`. List pause candidates. Do not
-  call `set_ad_entity_status` here — `omneky-launch-and-connect`
-  after one yes per entity set.
-- **audience-targeting-evidence.** Breakdowns by the dimensions
-  they named. Token lookup is `search_ad_targeting`. Applying
-  `update_ad_targeting` is launch, after a yes.
-- **target-cpa-budget-plan.** `minimum_budget_for_objective` +
-  `reach_estimate` + current `get_channel_budget`. Recommend a
-  daily number at or above the floor. Applying it is launch.
-
-## HubSpot (attribution caveats)
-
-`get_hubspot_connection_status` first. Disconnected → skip.
-
-Campaign-to-deal: `hubspot_search_deals` then `hubspot_get_deal`.
-There is **no join key** from a paid `campaign_id` to a HubSpot
-deal on this surface. Say that. You may place their named
-campaign/deal labels side by side and label the pairing
-**inferred, not attributed**. Do not claim paid-media ROAS from
-HubSpot. Confidence = low unless they supplied the join.
-
-<!-- host:begin name="abm-creative-handoff" -->
-ABM brief: `hubspot_search_companies` / `hubspot_get_company`
-(industry, name). Hand cited firmographic lines as
-`approved_claims` to a creative skill. Do not invent industry
-stats.
-<!-- host:end -->
-
-## Call by name (catalogue)
-
-Only when `list_brand_products` / `get_product_details` /
-`identify_product_from_url` appear on this session's `tools/list`.
-If they do not, say that capability isn't available on this
-connector. Do not invent a catalogue.
-
-| User ask | Tool |
-|---|---|
-| list my products / show SKUs / what's in the catalogue | `list_brand_products` |
-| details / images for one product | `get_product_details` |
-| import this product URL / paste a PDP | `identify_product_from_url` first, then scrape → finalize |
-
-`list_brand_products` needs numeric `brand_id`. Then
-`get_product_details` for images and copy used by a later ad. Do not
-mention a second name for product details here.
-
-## Import (short)
-
-Only when the import tools appear on `tools/list`. If they do
-not, say that capability isn't available on this connector.
-
-Stay on the chosen numeric `brand_id`. Catalogue writes need one
-confirmation of the fields being saved; reads need none.
-
-1. `identify_product_from_url` with the pasted `url` before any scrape.
-   - `is_single_product` false (listing, category, or home): do not
-     scrape. Ask for one product page URL.
-   - `true` or `null` (uncertain, often a JS-rendered page): you may
-     scrape. On `null`, say the page was uncertain.
-2. `scrape_product_from_url` with `brand_id` and that `url` (`locale`
-   defaults to `en-us`). It captures the first product on the page only.
-   - `status=pending`: retry **later** with the same `brand_id` and
-     `url`. Do not tight-loop this turn. Do not start a second scrape.
-   - `status=scraped`: show the fields and `scraped_image_urls`. The
-     user confirms which images and any edits.
-3. `finalize_scraped_product`. Required: `brand_id`, the scrape
-   `product_id`, `image_urls` (at least one), `product_name`, and
-   `product_description`. Zero images does not save. Do not finish a
-   scrape with `upsert_brand_product` (the draft stays hidden). Use the
-   **returned** catalogue `product_id` after this call.
-4. No URL, hand-built: `create_product`. Required: `brand_id`,
-   `product_name`, `product_description`. Confirm those fields first.
-   `create_product` does not accept image or video bytes. Catalogue
-   products cannot take uploaded media. If
-   `list_brand_products` is empty and another tool needs a
-   `product_id`, tell the user "I'll create a product you can reuse
-   later", create it, and use the new id without making the user
-   restart.
-   <!-- host:begin name="catalogue-creative-asset" -->
-   Pass a public HTTPS image URL on the original creative
-   call when that tool accepts one.
-   <!-- host:end -->
-5. Edit a completed product: `upsert_brand_product` with `product_id`,
-   and send the current `product_name` and `product_description` even
-   when they are unchanged. Confirm first. URL only:
-   `update_product_url` with `brand_id`, `product_id`, and `url`.
-   Confirm the URL. There is no product-delete tool.
-<!-- host:begin name="catalogue-creative-handoff" -->
-6. Hand off to `omneky-image-ads` or `omneky-product-video` for the
-   ad. A product image ad sends
-   **one** confirmed HTTPS image in `user_assets` (more only when the
-   user asks). Quality drops with more than two.
-<!-- host:end -->
-
-## If a catalogue tool is not available
-
-If this session cannot call `list_brand_products`,
-`get_product_details`, `identify_product_from_url`, or the other
-catalogue names, say that capability isn't available on this
-connector. Do **not** invent a catalogue or claim Omneky has no
-product list. The user can manage products in the Omneky app
-(Ads Central).
-
-## Connector reads
-
-Connector reads stay available but are not the hero path. Check status
-before a live read, and skip when disconnected:
-
-- HubSpot: `get_hubspot_connection_status`, then
-  `hubspot_search_contacts`, `hubspot_search_companies`, or
-  `hubspot_search_deals` (page size 10, cap 100). One record after
-  search: `hubspot_get_contact`, `hubspot_get_company`, or
-  `hubspot_get_deal` by id. Skip when disconnected.
-- Semrush: `get_semrush_connection_status`, then
-  `get_semrush_domain_overview` (`domain`) or
-  `get_semrush_keyword_overview` (`keyword`)
-- Ahrefs: `ahrefs_connection_status`, then `ahrefs_domain_rating`
-  (`target` is the domain).
-  Organic keywords: `ahrefs_organic_keywords` (default `limit` 10;
-  refuse `limit` greater than 50 — do not clamp). Backlinks:
-  `ahrefs_backlinks_stats` is an overview only, not a full export.
-- Search Console: `get_gsc_connection_status`, `gsc_list_sites`, then
-  `gsc_query_analytics` (`site_url`, `start_date`, `end_date`)
-
-These do not answer "which ads should I make more of."
-
 ## Failure and recovery
 
 Auth / 401 → `omneky-getting-started-and-failures` (host OAuth).
-Empty reporting → empty is missing data, not zero; narrow the window.
-`credit_insufficient` → nothing charged; the user can add credits in the Omneky app.
-`needs_user_decision` → `request_user_decision`; never dump JSON.
-Missing tool → say so; Omneky app. Do not invent it.
+Empty reporting → `get_performance_summary` (`include_coverage=true`); empty is not zero.
+`credit_insufficient` → nothing charged; state balance vs required
+(`omneky-billing`). No payment links.
+`needs_user_decision` → fill the missing detail and call again; ask in plain chat only if you cannot. Never dump JSON.
+Missing tool → this session may run an older published app snapshot: look
+the name up in [tool-name-fallbacks](../_shared/references/tool-name-fallbacks.md)
+and call the older name if that one is on `tools/list`. If neither is
+listed, say so; Omneky app. Do not invent it.
 Never ask for internal ids. Never silent-switch `brand_id`.
 
 <!-- host:begin name="capability-gaps-link" -->
 ## Gaps
 
-This connector cannot do comment/social inbox, landing pages,
-subtitle burn-in, email/SMS, organic scheduling, Meta Ad Library
-competitor ads, virtual try-on, or website builder. Decline those.
-See [capability-gaps](../_shared/references/capability-gaps.md).
+Decline social inbox, landing pages, subtitle burn-in, email/SMS,
+organic scheduling, Ad Library ads, try-on, and website builder
+([capability-gaps](../_shared/references/capability-gaps.md)).
 <!-- host:end -->
Full technical diff · 3 changed fields

changed /description

BEFORE
"Call list_brand_products, identify_product_from_url, and get_product_details by name (list my products, import this product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly performance report, HubSpot, and read-only budget reallocation. Do not use for first-run onboarding (omneky-onboarding), agency cross-brand (omneky-agency), competitor positioning (omneky-competitor-analysis), A/B tests (omneky-ab-tests), weekly marketing plans (omneky-marketing-plan), generating creatives, launch (omneky-launch-and-connect), or credit balance (omneky-billing). Keywords: list my products, import this URL, catalogue, SKU, PDP, which ads should I make more of, winning ads, ROAS drop, CPA drop, CTR, weekly performance report, fatigue refresh, budget reallocation, waste pruner, audience targeting, target CPA, HubSpot attribution, ABM, Semrush, Ahrefs, GSC."
AFTER
"Call get_products, import_product_from_url, and get_products by name (list my products, import this product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly performance report, HubSpot, and read-only budget reallocation. Do not use for first-run onboarding (omneky-onboarding), agency cross-brand (omneky-agency), competitor positioning (omneky-competitor-analysis), A/B tests (omneky-ab-tests), weekly marketing plans (omneky-marketing-plan), generating creatives, launch (omneky-launch-and-connect), or credit balance (omneky-billing). Keywords: list my products, import this URL, catalogue, SKU, PDP, which ads should I make more of, winning ads, ROAS drop, CPA drop, CTR, weekly performance report, fatigue refresh, budget reallocation, waste pruner, audience targeting, target CPA, HubSpot attribution, ABM, Semrush, Ahrefs."

changed /included_files

BEFORE
[]
AFTER
[
  {
    "relative_path": "references/playbook.md",
    "size_in_bytes": 13193
  }
]

changed /skill_md_contents

BEFORE
"---\nname: omneky-analytics-and-catalogue\ndescription: >-\n  Call list_brand_products, identify_product_from_url, and\n  get_product_details by name (list my products, import this\n  product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly\n  performance report, HubSpot, and read-only budget\n  reallocation. Do not use for first-run onboarding\n  (omneky-onboarding), agency cross-brand (omneky-agency),\n  competitor positioning (omneky-competitor-analysis), A/B\n  tests (omneky-ab-tests), weekly marketing plans\n  (omneky-marketing-plan), generating creatives, launch\n  (omneky-launch-and-connect), or credit balance\n  (omneky-billing). Keywords: list my products, import this URL,\n  catalogue, SKU, PDP, which ads should I make more of, winning\n  ads, ROAS drop, CPA drop, CTR, weekly performance report,\n  fatigue refresh, budget reallocation, waste pruner, audience\n  targeting, target CPA, HubSpot attribution, ABM, Semrush,\n  Ahrefs, GSC.\n---\n\n# Ad performance and product catalogue\n\nThe connected Omneky MCP server exposes the names below. Call them\n**by those exact names** when they appear on this session's\n`tools/list`. Metrics and reporting reads work without catalogue\ntools. Catalogue names (`list_brand_products`, `get_product_details`,\n`identify_product_from_url`, and the import / create / upsert tools)\nare optional. If they are not listed, say that capability isn't\navailable on this connector. Do not invent a catalogue or claim\nOmneky has no product list.\n\nThis skill is analysis and catalogue only. Competitor\npositioning → `omneky-competitor-analysis`.\n<!-- host:begin name=\"ab-tests-routing\" -->\nA/B tests → `omneky-ab-tests`.\n<!-- host:end -->\nWeekly plans → `omneky-marketing-plan`.\n<!-- host:begin name=\"creative-skill-routing\" -->\nImage ads → `omneky-image-ads`. Product video →\n`omneky-product-video`. Edit /\nresize → `omneky-edit-and-resize`. Launch →\n`omneky-launch-and-connect`. Credits → `omneky-billing`.\n<!-- host:end -->\n\nHandoff fields:\n[handoff-contract](../_shared/references/handoff-contract.md).\nSourced ids (how to obtain / what to do if absent):\n[preconditions](../_shared/references/preconditions.md).\n<!-- host:begin name=\"scan-tools-routing\" -->\nAcceptance gates (for a later creative):\n[creative-quality-gates](../_shared/references/creative-quality-gates.md).\nStructured brief (for a later creative):\n[creative-brief](../_shared/references/creative-brief.md).\nOther intents:\n[capability-registry](../_shared/references/capability-registry.md).\nWrite for:\n[marketer-persona](../_shared/references/marketer-persona.md).\n<!-- host:end -->\n\n## Goal\n\nRanked evidence, catalogue records, and connector reads. No render. No spend mutation. Keep the weekly performance report.\n\n## Prerequisites\n\n| Need | Check | If missing |\n|---|---|---|\n| Brand | `list_brands` | Ask which; do not invent |\n| Coverage | Metric tool for the stated window | Empty is missing data; do not report zeros |\n| Winners / diagnosis | `get_performance_breakdown` / `get_daily_metrics` / `get_performance_movers` | Narrow the window |\n| Product | `list_brand_products` when listed | Import via `identify_product_from_url` → scrape → finalize. If catalogue tools are not on `tools/list`, say that capability isn't available on this connector |\n| HubSpot | `get_hubspot_connection_status` | Skip CRM joins; say disconnected |\n| Semrush / Ahrefs / GSC | matching status tool | Skip that source |\n| Budget apply | (read-only here) | `omneky-launch-and-connect` after a yes (`set_ad_budget` / `set_ad_entity_status`) |\n\n## Flow\n\nLead with the answer (ROAS, CPA, CTR, spend, lift), one recommendation, and confidence. Recipes: winners, fatigue, weekly performance report, ROAS/CPA drop, waste pruner (evidence), targeting evidence, target-CPA plan, HubSpot attribution / ABM brief, catalogue import. Do not open a separate memo skill.\n\nCollect `source_evidence` from the metric call you just ran: metric,\nwindow (start and end dates, calendar or rolling), channel, ad or\ncreative ids, and sample size when handing a performance-informed\nbrief to creative or a plan-driven launch. Collect `approved_claims`\nfrom the user or the product page / catalogue copy they pointed at;\neach claim needs its source. Do not invent a claim. Plain\nuser-directed edits and launches do not need `source_evidence`.\n\n## When to use\n\n- \"Which ads should I make more of?\", winning ads, next batch brief\n- ROAS, CTR, CPA, spend, movers, daily paid-media charts\n- List products / SKUs, or import a product URL before an ad\n  (only when those catalogue tools appear on `tools/list`)\n\n## When not to use\n\n- Competitor positioning / messaging gaps\n  (`omneky-competitor-analysis`)\n<!-- host:begin name=\"ab-tests-when-not\" -->\n- A/B test create or \"which variant won\" (`omneky-ab-tests`)\n<!-- host:end -->\n- Weekly marketing plan create / revise / poll / approve\n  (`omneky-marketing-plan`)\n<!-- host:begin name=\"creative-when-not\" -->\n- Generating the image or video (hand off to `omneky-image-ads` or\n  `omneky-product-video`)\n<!-- host:end -->\n- Launching or connecting campaigns (`omneky-launch-and-connect`)\n- Plan, credit history, or card on file (`omneky-billing`)\n<!-- host:begin name=\"onboarding-agency-when-not\" -->\n- First-run setup / first ad today (`omneky-onboarding`)\n- Cross-brand / all clients (`omneky-agency`)\n<!-- host:end -->\n- No product-delete tool\n\n## Winning ads → next batch\n\nStay on the chosen brand for every call below. Empty rows are missing\ncoverage, not a reason to silent-switch brands.\n\n1. Reuse the `brand_id` already chosen in this conversation. Unknown:\n   `list_brands` (pass `search` when the user named the brand). Several\n   matches: ask which one and wait. One clear match: say the name and\n   id, then keep that `brand_id`.\n2. Pick the date window before any metric call.\n   - User named dates: use those inclusive `YYYY-MM-DD` days.\n   - \"This week\" means the calendar week to date, not the last 7\n     days. Use last 7 days only when the user says \"last 7 days\".\n     Movers keep `period_days` (default 7) and do not take\n     `date_range`. A winner list with no dates uses the last 30\n     days. Always say which window you used.\n   - This surface has no latest-date tool. Do not invent one.\n3. Coverage: an empty metric result means there is no imported\n   data for that brand, window, or channel.\n   - Do not report zeros. Narrow to a shorter or more recent window,\n     or stop and say there is no imported data.\n   - Empty metrics are missing data, not a zero ROAS, spend, or CTR.\n4. Then, always passing that `brand_id` (it is optional in some\n   schemas; send it anyway so results stay on the chosen brand):\n   - Ranked winners: `get_performance_breakdown` with `dimension`\n     (creative, campaign, ad, or channel), `date_range` as two inclusive\n     dates `[\"YYYY-MM-DD\", \"YYYY-MM-DD\"]`, `sort_metric` (the metric to\n     rank by, such as `spend` or `roas`; the schema does not list\n     values), `sort_order` (`desc` for winners), and\n     `results_per_page` (10 is enough for a winner list). Pass\n     `filters` as `[]` when not filtering. There is no `metrics`\n     argument; set `include_all_metrics` to `true` when you need every\n     column.\n   - Day-by-day: `get_daily_metrics` with `date_range` (same shape) and\n     `filters` as `[]` when not filtering. There is no `metrics`\n     argument.\n   - What changed vs the prior period: `get_performance_movers` with\n     `dimension`, `metric`, `end_date` (last day of the current window,\n     `YYYY-MM-DD`), and `period_days` (default 7). It builds the prior\n     window. Do not send `date_range` to this tool.\n   - Name the user said: `search_reporting_values` with `dimension` and\n     `q` (the text). It returns labels, not metrics. Put the exact value\n     in a later filter. Operators are `includes` or `excludes` only\n     (`condition`, `dimension`, `value` as a list of strings). Do not\n     invent `=`, `in`, or `gt`.\n   - Current allocated daily budget (not spend, not historical ROAS):\n     `get_channel_budget` with `channels`. Totals plus per-campaign\n     `breakdown` / `basis` when the backend returns them. A disconnected\n     or skipped channel is a per-channel error, not an empty total.\n   - Fill any other required field from the tool's own schema and\n     earlier results in this session. Do not ask the user for internal\n     ids.\n5. Never sum `selected_conversion_metric_value` across channels. That\n   field is the brand's goal and it differs per channel (purchases on\n   one channel, conversions on another). Group by channel, or use a\n   named metric (`purchases`, clicks, spend, ROAS, CTR).\n6. Minimum evidence (also in the handoff contract). State the\n   sample next to every conclusion. Fewer than about 1,000\n   impressions, 30 clicks, or 10 conversions per ad is thin — say\n   \"too little data to conclude\" and show the numbers. Attribution\n   differs by channel; do not mix attributed totals or sum\n   `selected_conversion_metric_value` across channels. An aggregate\n   ratio is summed numerators over summed denominators, never the\n   mean of per-row ratios.\n7. `get_recommendations` when the user asks what to try next. This is\n   advice, not the metric table from steps 4–5.\n   - Required: `brand_id`.\n   - `start_date` and `end_date` are inclusive `YYYY-MM-DD`. Omit both\n     for the last 30 days. A window longer than 30 days is capped at\n     30. `metric` defaults to `spend`.\n   - `status=timeout`: narrow the dates. Do not retry the identical\n     call in this turn.\n<!-- host:begin name=\"make-next-batch\" -->\n8. To **make** the next batch, hand the winning product or concept to\n   `omneky-image-ads` or `omneky-product-video` with the handoff\n   contract (brand, product, source evidence, approved claims,\n   `source_ad_ids`, and one hypothesis per requested variant). This\n   skill does not generate creatives. Launch is a later handoff. A\n   weekly plan is `omneky-marketing-plan`, not a reconstructed\n   generate/launch sequence here.\n<!-- host:end -->\n\n<!-- host:begin name=\"performance-to-creative\" -->\n## Performance-to-creative flows\n\nEvidence first, then a handoff. Do not invent ROAS. Do not generate\nhere. Each requested variant is one hypothesis: source ad, what\nstays constant, what changes, audience / placement, success metric.\n\n- **fatigue-refresh.** `get_performance_movers` on Meta (falling\n  CTR or ROAS; state the window and sample). Pick up to 3 fatigued\n  ads with enough evidence. Hand 3 replacement hypotheses to\n  `omneky-image-ads` (or `omneky-product-video` when they asked for\n  video).\n- **top-roas-variations.** `get_performance_breakdown` sorted by\n  `roas`. Take the verified winners. Hand N distinct hypotheses\n  (N = the count they asked for) to `omneky-image-ads`.\n- **paused-campaign-from-winners.** After the creative skill\n  returns accepted assets, hand `omneky-launch-and-connect` with\n  those URLs. Launch stays `PAUSED` unless they explicitly say to\n  go live.\n- **weekly-performance-report.** Same coverage check, then\n  `get_performance_breakdown` + `get_daily_metrics` +\n  `get_performance_movers`. State the window. Thin samples stay\n  \"too little data to conclude.\" Do not declare an A/B winner\n  here (`omneky-ab-tests`).\n- **budget-reallocation.** Read `get_channel_budget` and the\n  winner / mover tables. Recommend a move (from / to / amount /\n  why). This skill is read-only. Applying `set_ad_budget` is\n  `omneky-launch-and-connect` after one yes. Do not mutate spend\n  here.\n\nThin samples stay suggestions: say \"too little data to conclude\"\nand still show the numbers.\n<!-- host:end -->\n\n- **roas-cpa-drop-diagnosis.** Coverage first. Then\n  `get_performance_movers` (falling ROAS or CPA),\n  `get_performance_breakdown`, and `get_daily_metrics`. Explain\n  with evidence only (window, sample, channel). Confidence is\n  thin when samples are thin.\n- **waste-pruner (evidence).** Rank by spend with weak CTR / ROAS\n  on `get_performance_breakdown`. List pause candidates. Do not\n  call `set_ad_entity_status` here — `omneky-launch-and-connect`\n  after one yes per entity set.\n- **audience-targeting-evidence.** Breakdowns by the dimensions\n  they named. Token lookup is `search_ad_targeting`. Applying\n  `update_ad_targeting` is launch, after a yes.\n- **target-cpa-budget-plan.** `minimum_budget_for_objective` +\n  `reach_estimate` + current `get_channel_budget`. Recommend a\n  daily number at or above the floor. Applying it is launch.\n\n## HubSpot (attribution caveats)\n\n`get_hubspot_connection_status` first. Disconnected → skip.\n\nCampaign-to-deal: `hubspot_search_deals` then `hubspot_get_deal`.\nThere is **no join key** from a paid `campaign_id` to a HubSpot\ndeal on this surface. Say that. You may place their named\ncampaign/deal labels side by side and label the pairing\n**inferred, not attributed**. Do not claim paid-media ROAS from\nHubSpot. Confidence = low unless they supplied the join.\n\n<!-- host:begin name=\"abm-creative-handoff\" -->\nABM brief: `hubspot_search_companies` / `hubspot_get_company`\n(industry, name). Hand cited firmographic lines as\n`approved_claims` to a creative skill. Do not invent industry\nstats.\n<!-- host:end -->\n\n## Call by name (catalogue)\n\nOnly when `list_brand_products` / `get_product_details` /\n`identify_product_from_url` appear on this session's `tools/list`.\nIf they do not, say that capability isn't available on this\nconnector. Do not invent a catalogue.\n\n| User ask | Tool |\n|---|---|\n| list my products / show SKUs / what's in the catalogue | `list_brand_products` |\n| details / images for one product | `get_product_details` |\n| import this product URL / paste a PDP | `identify_product_from_url` first, then scrape → finalize |\n\n`list_brand_products` needs numeric `brand_id`. Then\n`get_product_details` for images and copy used by a later ad. Do not\nmention a second name for product details here.\n\n## Import (short)\n\nOnly when the import tools appear on `tools/list`. If they do\nnot, say that capability isn't available on this connector.\n\nStay on the chosen numeric `brand_id`. Catalogue writes need one\nconfirmation of the fields being saved; reads need none.\n\n1. `identify_product_from_url` with the pasted `url` before any scrape.\n   - `is_single_product` false (listing, category, or home): do not\n     scrape. Ask for one product page URL.\n   - `true` or `null` (uncertain, often a JS-rendered page): you may\n     scrape. On `null`, say the page was uncertain.\n2. `scrape_product_from_url` with `brand_id` and that `url` (`locale`\n   defaults to `en-us`). It captures the first product on the page only.\n   - `status=pending`: retry **later** with the same `brand_id` and\n     `url`. Do not tight-loop this turn. Do not start a second scrape.\n   - `status=scraped`: show the fields and `scraped_image_urls`. The\n     user confirms which images and any edits.\n3. `finalize_scraped_product`. Required: `brand_id`, the scrape\n   `product_id`, `image_urls` (at least one), `product_name`, and\n   `product_description`. Zero images does not save. Do not finish a\n   scrape with `upsert_brand_product` (the draft stays hidden). Use the\n   **returned** catalogue `product_id` after this call.\n4. No URL, hand-built: `create_product`. Required: `brand_id`,\n   `product_name`, `product_description`. Confirm those fields first.\n   `create_product` does not accept image or video bytes. Catalogue\n   products cannot take uploaded media. If\n   `list_brand_products` is empty and another tool needs a\n   `product_id`, tell the user \"I'll create a product you can reuse\n   later\", create it, and use the new id without making the user\n   restart.\n   <!-- host:begin name=\"catalogue-creative-asset\" -->\n   Pass a public HTTPS image URL on the original creative\n   call when that tool accepts one.\n   <!-- host:end -->\n5. Edit a completed product: `upsert_brand_product` with `product_id`,\n   and send the current `product_name` and `product_description` even\n   when they are unchanged. Confirm first. URL only:\n   `update_product_url` with `brand_id`, `product_id`, and `url`.\n   Confirm the URL. There is no product-delete tool.\n<!-- host:begin name=\"catalogue-creative-handoff\" -->\n6. Hand off to `omneky-image-ads` or `omneky-product-video` for the\n   ad. A product image ad sends\n   **one** confirmed HTTPS image in `user_assets` (more only when the\n   user asks). Quality drops with more than two.\n<!-- host:end -->\n\n## If a catalogue tool is not available\n\nIf this session cannot call `list_brand_products`,\n`get_product_details`, `identify_product_from_url`, or the other\ncatalogue names, say that capability isn't available on this\nconnector. Do **not** invent a catalogue or claim Omneky has no\nproduct list. The user can manage products in the Omneky app\n(Ads Central).\n\n## Connector reads\n\nConnector reads stay available but are not the hero path. Check status\nbefore a live read, and skip when disconnected:\n\n- HubSpot: `get_hubspot_connection_status`, then\n  `hubspot_search_contacts`, `hubspot_search_companies`, or\n  `hubspot_search_deals` (page size 10, cap 100). One record after\n  search: `hubspot_get_contact`, `hubspot_get_company`, or\n  `hubspot_get_deal` by id. Skip when disconnected.\n- Semrush: `get_semrush_connection_status`, then\n  `get_semrush_domain_overview` (`domain`) or\n  `get_semrush_keyword_overview` (`keyword`)\n- Ahrefs: `ahrefs_connection_status`, then `ahrefs_domain_rating`\n  (`target` is the domain).\n  Organic keywords: `ahrefs_organic_keywords` (default `limit` 10;\n  refuse `limit` greater than 50 — do not clamp). Backlinks:\n  `ahrefs_backlinks_stats` is an overview only, not a full export.\n- Search Console: `get_gsc_connection_status`, `gsc_list_sites`, then\n  `gsc_query_analytics` (`site_url`, `start_date`, `end_date`)\n\nThese do not answer \"which ads should I make more of.\"\n\n## Failure and recovery\n\nAuth / 401 → `omneky-getting-started-and-failures` (host OAuth).\nEmpty reporting → empty is missing data, not zero; narrow the window.\n`credit_insufficient` → nothing charged; the user can add credits in the Omneky app.\n`needs_user_decision` → `request_user_decision`; never dump JSON.\nMissing tool → say so; Omneky app. Do not invent it.\nNever ask for internal ids. Never silent-switch `brand_id`.\n\n<!-- host:begin name=\"capability-gaps-link\" -->\n## Gaps\n\nThis connector cannot do comment/social inbox, landing pages,\nsubtitle burn-in, email/SMS, organic scheduling, Meta Ad Library\ncompetitor ads, virtual try-on, or website builder. Decline those.\nSee [capability-gaps](../_shared/references/capability-gaps.md).\n<!-- host:end -->\n"
AFTER
"---\nname: omneky-analytics-and-catalogue\ndescription: >-\n  Call get_products, import_product_from_url, and\n  get_products by name (list my products, import this\n  product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly\n  performance report, HubSpot, and read-only budget\n  reallocation. Do not use for first-run onboarding\n  (omneky-onboarding), agency cross-brand (omneky-agency),\n  competitor positioning (omneky-competitor-analysis), A/B\n  tests (omneky-ab-tests), weekly marketing plans\n  (omneky-marketing-plan), generating creatives, launch\n  (omneky-launch-and-connect), or credit balance\n  (omneky-billing). Keywords: list my products, import this URL,\n  catalogue, SKU, PDP, which ads should I make more of, winning\n  ads, ROAS drop, CPA drop, CTR, weekly performance report,\n  fatigue refresh, budget reallocation, waste pruner, audience\n  targeting, target CPA, HubSpot attribution, ABM, Semrush,\n  Ahrefs.\n---\n\n# Ad performance and product catalogue\n\nUse this when the user asks how their ads perform, which ads to make\nmore of, or wants to list or import products. One job: ranked\nevidence and catalogue records. No render, no spend change.\n\nCall tools by their exact names when they are on `tools/list`.\nCatalogue tools are optional; if they are missing, say that\ncapability isn't available on this connector.\n\nHandoff fields: [handoff-contract](../_shared/references/handoff-contract.md).\nSourced ids (how to obtain / what to do if absent): [preconditions](../_shared/references/preconditions.md).\n<!-- host:begin name=\"scan-tools-routing\" -->\nAcceptance gates (for a later creative): [creative-quality-gates](../_shared/references/creative-quality-gates.md).\nStructured brief (for a later creative): [creative-brief](../_shared/references/creative-brief.md).\nOther intents: [capability-registry](../_shared/references/capability-registry.md).\nWrite for: [marketer-persona](../_shared/references/marketer-persona.md).\n<!-- host:end -->\n\n## Goal\n\nRanked evidence, catalogue records, and connector reads.\n\n## Prerequisites\n\n| Need | Check | If missing |\n|---|---|---|\n| Brand | `list_brands` | Ask which; do not invent |\n| Coverage | `get_performance_summary` (`include_coverage=true`) | Say there is no imported data; do not report zeros |\n| Winners / diagnosis | `get_performance_breakdown` / `get_performance_summary` / `get_performance_movers` | Narrow the window |\n| Product | `get_products` when listed | Import via `import_product_from_url` (`url`), then `save_imported_product` (`draft_id`). If catalogue tools are not on `tools/list`, say that capability isn't available on this connector |\n| HubSpot | `list_connections` (`connector=\"hubspot\"`) | Skip CRM joins; say disconnected |\n| Semrush / Ahrefs | matching status tool | Skip that source |\n| Budget apply | (read-only here) | `omneky-launch-and-connect` (`set_ad_budget` / `set_ad_status`; the host confirms) |\n\n## Flow\n\n1. Resolve the brand and confirm coverage for the window.\n2. Run the recipe the user asked for: winners, fatigue, weekly\n   report, ROAS/CPA drop, waste pruner (evidence only), targeting\n   evidence, target-CPA plan, HubSpot attribution, ABM brief, or\n   catalogue import. Details: [playbook](references/playbook.md).\n3. Lead with the answer (ROAS, CPA, CTR, spend, lift), one\n   recommendation, and confidence.\n4. When handing off to creative or a plan-driven launch, attach\n   `source_evidence` (metric, window, channel, ad ids, sample size)\n   and `approved_claims`, each with its source. Never invent a claim.\n\n## When to use\n\n- \"Which ads should I make more of?\", winning ads, next batch brief\n- ROAS, CTR, CPA, spend, movers, daily paid-media charts\n- List products / SKUs, or import a product URL before an ad\n  (only when those catalogue tools appear on `tools/list`)\n\n## When not to use\n\n- Competitor positioning / messaging gaps\n  (`omneky-competitor-analysis`)\n<!-- host:begin name=\"ab-tests-when-not\" -->\n- A/B test create or \"which variant won\" (`omneky-ab-tests`)\n<!-- host:end -->\n- Weekly marketing plan create / revise / poll / approve\n  (`omneky-marketing-plan`)\n<!-- host:begin name=\"creative-when-not\" -->\n- Generating the image or video (hand off to `omneky-image-ads` or\n  `omneky-product-video`)\n<!-- host:end -->\n- Launching or connecting campaigns (`omneky-launch-and-connect`)\n- Credit balance or plan upgrade (`omneky-billing`)\n<!-- host:begin name=\"onboarding-agency-when-not\" -->\n- First-run setup / first ad today (`omneky-onboarding`)\n- Cross-brand / all clients (`omneky-agency`)\n<!-- host:end -->\n- No product-delete tool\n\n## Failure and recovery\n\nAuth / 401 → `omneky-getting-started-and-failures` (host OAuth).\nEmpty reporting → `get_performance_summary` (`include_coverage=true`); empty is not zero.\n`credit_insufficient` → nothing charged; state balance vs required\n(`omneky-billing`). No payment links.\n`needs_user_decision` → fill the missing detail and call again; ask in plain chat only if you cannot. Never dump JSON.\nMissing tool → this session may run an older published app snapshot: look\nthe name up in [tool-name-fallbacks](../_shared/references/tool-name-fallbacks.md)\nand call the older name if that one is on `tools/list`. If neither is\nlisted, say so; Omneky app. Do not invent it.\nNever ask for internal ids. Never silent-switch `brand_id`.\n\n<!-- host:begin name=\"capability-gaps-link\" -->\n## Gaps\n\nDecline social inbox, landing pages, subtitle burn-in, email/SMS,\norganic scheduling, Ad Library ads, try-on, and website builder\n([capability-gaps](../_shared/references/capability-gaps.md)).\n<!-- host:end -->\n"
Full snapshot data
{
  "description": "Call get_products, import_product_from_url, and get_products by name (list my products, import this product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly performance report, HubSpot, and read-only budget reallocation. Do not use for first-run onboarding (omneky-onboarding), agency cross-brand (omneky-agency), competitor positioning (omneky-competitor-analysis), A/B tests (omneky-ab-tests), weekly marketing plans (omneky-marketing-plan), generating creatives, launch (omneky-launch-and-connect), or credit balance (omneky-billing). Keywords: list my products, import this URL, catalogue, SKU, PDP, which ads should I make more of, winning ads, ROAS drop, CPA drop, CTR, weekly performance report, fatigue refresh, budget reallocation, waste pruner, audience targeting, target CPA, HubSpot attribution, ABM, Semrush, Ahrefs.",
  "included_files": [
    {
      "relative_path": "references/playbook.md",
      "size_in_bytes": 13193
    }
  ],
  "name": "omneky-analytics-and-catalogue",
  "skill_md_contents": "---\nname: omneky-analytics-and-catalogue\ndescription: >-\n  Call get_products, import_product_from_url, and\n  get_products by name (list my products, import this\n  product URL, SKU, PDP). Also winning ads, ROAS, CTR, weekly\n  performance report, HubSpot, and read-only budget\n  reallocation. Do not use for first-run onboarding\n  (omneky-onboarding), agency cross-brand (omneky-agency),\n  competitor positioning (omneky-competitor-analysis), A/B\n  tests (omneky-ab-tests), weekly marketing plans\n  (omneky-marketing-plan), generating creatives, launch\n  (omneky-launch-and-connect), or credit balance\n  (omneky-billing). Keywords: list my products, import this URL,\n  catalogue, SKU, PDP, which ads should I make more of, winning\n  ads, ROAS drop, CPA drop, CTR, weekly performance report,\n  fatigue refresh, budget reallocation, waste pruner, audience\n  targeting, target CPA, HubSpot attribution, ABM, Semrush,\n  Ahrefs.\n---\n\n# Ad performance and product catalogue\n\nUse this when the user asks how their ads perform, which ads to make\nmore of, or wants to list or import products. One job: ranked\nevidence and catalogue records. No render, no spend change.\n\nCall tools by their exact names when they are on `tools/list`.\nCatalogue tools are optional; if they are missing, say that\ncapability isn't available on this connector.\n\nHandoff fields: [handoff-contract](../_shared/references/handoff-contract.md).\nSourced ids (how to obtain / what to do if absent): [preconditions](../_shared/references/preconditions.md).\n<!-- host:begin name=\"scan-tools-routing\" -->\nAcceptance gates (for a later creative): [creative-quality-gates](../_shared/references/creative-quality-gates.md).\nStructured brief (for a later creative): [creative-brief](../_shared/references/creative-brief.md).\nOther intents: [capability-registry](../_shared/references/capability-registry.md).\nWrite for: [marketer-persona](../_shared/references/marketer-persona.md).\n<!-- host:end -->\n\n## Goal\n\nRanked evidence, catalogue records, and connector reads.\n\n## Prerequisites\n\n| Need | Check | If missing |\n|---|---|---|\n| Brand | `list_brands` | Ask which; do not invent |\n| Coverage | `get_performance_summary` (`include_coverage=true`) | Say there is no imported data; do not report zeros |\n| Winners / diagnosis | `get_performance_breakdown` / `get_performance_summary` / `get_performance_movers` | Narrow the window |\n| Product | `get_products` when listed | Import via `import_product_from_url` (`url`), then `save_imported_product` (`draft_id`). If catalogue tools are not on `tools/list`, say that capability isn't available on this connector |\n| HubSpot | `list_connections` (`connector=\"hubspot\"`) | Skip CRM joins; say disconnected |\n| Semrush / Ahrefs | matching status tool | Skip that source |\n| Budget apply | (read-only here) | `omneky-launch-and-connect` (`set_ad_budget` / `set_ad_status`; the host confirms) |\n\n## Flow\n\n1. Resolve the brand and confirm coverage for the window.\n2. Run the recipe the user asked for: winners, fatigue, weekly\n   report, ROAS/CPA drop, waste pruner (evidence only), targeting\n   evidence, target-CPA plan, HubSpot attribution, ABM brief, or\n   catalogue import. Details: [playbook](references/playbook.md).\n3. Lead with the answer (ROAS, CPA, CTR, spend, lift), one\n   recommendation, and confidence.\n4. When handing off to creative or a plan-driven launch, attach\n   `source_evidence` (metric, window, channel, ad ids, sample size)\n   and `approved_claims`, each with its source. Never invent a claim.\n\n## When to use\n\n- \"Which ads should I make more of?\", winning ads, next batch brief\n- ROAS, CTR, CPA, spend, movers, daily paid-media charts\n- List products / SKUs, or import a product URL before an ad\n  (only when those catalogue tools appear on `tools/list`)\n\n## When not to use\n\n- Competitor positioning / messaging gaps\n  (`omneky-competitor-analysis`)\n<!-- host:begin name=\"ab-tests-when-not\" -->\n- A/B test create or \"which variant won\" (`omneky-ab-tests`)\n<!-- host:end -->\n- Weekly marketing plan create / revise / poll / approve\n  (`omneky-marketing-plan`)\n<!-- host:begin name=\"creative-when-not\" -->\n- Generating the image or video (hand off to `omneky-image-ads` or\n  `omneky-product-video`)\n<!-- host:end -->\n- Launching or connecting campaigns (`omneky-launch-and-connect`)\n- Credit balance or plan upgrade (`omneky-billing`)\n<!-- host:begin name=\"onboarding-agency-when-not\" -->\n- First-run setup / first ad today (`omneky-onboarding`)\n- Cross-brand / all clients (`omneky-agency`)\n<!-- host:end -->\n- No product-delete tool\n\n## Failure and recovery\n\nAuth / 401 → `omneky-getting-started-and-failures` (host OAuth).\nEmpty reporting → `get_performance_summary` (`include_coverage=true`); empty is not zero.\n`credit_insufficient` → nothing charged; state balance vs required\n(`omneky-billing`). No payment links.\n`needs_user_decision` → fill the missing detail and call again; ask in plain chat only if you cannot. Never dump JSON.\nMissing tool → this session may run an older published app snapshot: look\nthe name up in [tool-name-fallbacks](../_shared/references/tool-name-fallbacks.md)\nand call the older name if that one is on `tools/list`. If neither is\nlisted, say so; Omneky app. Do not invent it.\nNever ask for internal ids. Never silent-switch `brand_id`.\n\n<!-- host:begin name=\"capability-gaps-link\" -->\n## Gaps\n\nDecline social inbox, landing pages, subtitle burn-in, email/SMS,\norganic scheduling, Ad Library ads, try-on, and website builder\n([capability-gaps](../_shared/references/capability-gaps.md)).\n<!-- host:end -->\n"
}

SHA-256 of public snapshot: 951ea7365ab320d143fe1a6701a973291b19fa9bf516f57d00c00c3d717e68a8