القائمة
Use when launching a Google Ads experiment. Loaded after the create master skill.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use when researching competitor ad creatives: find competitors, brand IDs, evergreen winners, breakout winners, pattern analysis.
Use when generating Word documents (.docx). Uses the marble_docx API.
Use when generating PDF documents. Uses the marble_pdf API.
Use when generating PowerPoint presentations (.pptx). Uses the marble_pptx API.
Use when generating Excel workbooks (.xlsx). Pandas + openpyxl patterns for multiple sheets, number formats, freeze panes, formulas.
Use for any GA4 query whose result will be reported back to the user. Covers conversions ≠ transactions distinction, channel-subset-sum trap with sessionDefaultChannelGroup, intraday partial-data discipline, Paid Social aggregation rules, API-vs-UI attribution mismatch.
| name | google-ads-create-experiment |
| description | Use when launching a Google Ads experiment. Loaded after the create master skill. |
Tools: google_ads_propose_create_experiment, google_ads_propose_update_experiment. Tool schemas describe shape; this skill covers the business rules.
type: SEARCH_CUSTOM only. No DISPLAY_CUSTOM, no SHOPPING, no YOUTUBE. Reject any other type up front.base_campaign_id: must reference an existing campaign with advertising_channel_type = SEARCH.Two-arm A/B test on a single base campaign:
SETUP), then both ExperimentArms — control reuses base campaign; treatment forks a draft campaign.draft_modifications apply to the treatment's draft campaign.experiments/{id}:scheduleExperiment (Google long-running operation) and polls until done.traffic_split.control + traffic_split.treatment must sum to 100. Each in 1–99.start_date: ≥ today + 1 day, within ~1 year.end_date: > start_date.name is unique per customer — duplicate names reject. Append a suffix and retry.draft_modifications.daily_budget is propose-side advertised but the executor does NOT push it. Tell the user budget changes on the draft campaign aren't supported in V1; propose a separate post-schedule update if they want one.We expose ENABLE/PAUSE-style transitions and Google's lifecycle promotions. We do not expose entity removal.
action | Effect | Allowed from |
|---|---|---|
RENAME | Rename experiment | any state |
PAUSE | Pause running experiment | RUNNING |
RESUME | Resume paused experiment | HALTED |
END | End experiment immediately (LRO) | RUNNING / GRADUATED |
PROMOTE | Promote treatment to base (LRO) | RUNNING / ENDED |
GRADUATE | Convert treatment into a standalone campaign (LRO) | RUNNING |
END, PROMOTE, GRADUATE use Google's long-running operation system. The executor polls until done; failures show up in the LRO's error field. Common LRO errors:
START_DATE_TOO_FAR_IN_FUTURECANNOT_CREATE_EXPERIMENT_CAMPAIGN_WITH_SHARED_BUDGETSurface these directly to the user — they're fixable in the Ads UI by adjusting dates or detaching the shared budget.
Create:
Done. Created and scheduled experiment
<name>(id<id>).<control>/<treatment>split on campaign<base_name>. Treatment draft:<draft_name>. Will start<start_date>.
Update:
Done.
<action>on experiment<name>(id<id>).