Prepare proposed paid-ad changes for review, apply only the changes the marketer explicitly approves, and record decisions and outcomes. This is an internal step of the run-paid-campaigns playbook and is intended to be used through that workflow.
Step of the run-paid-campaigns playbook: stages the week's approval items, applies only what the marketer approves and keeps the change log. Run run-paid-campaigns instead; not for direct use.
Run score Not tested
See how to installReview the instructions the agent will receive before you install.
Load: ../run-paid-campaigns/references/meta-levers.md, ../run-paid-campaigns/references/proposal-and-log.md (exact grammar). Steps by each move's lever (the playbook's Before you start 3): GooseWorks levers — 1–8; connector levers — 1, 2, 4, then 9–11; written instructions — 1, 2, 4 (no tool call), then 7 on the marketer's "done". Local mode has only the last two and calls no gooseworks: tool; its files live under gooseworks-local/<brand-slug>/ (host file tools).
ads/campaigns/<slug>/state.md. If it doesn't exist, create it with the frontmatter (campaign, stage, goose_campaign_id, meta_push_ids: [], meta_campaign_id, updated_at, updated_by; local mode: no ids), the header lines, ## Open recommendations and ## Meta objects.- R<n> (YYYY-MM-DD): <move> — evidence <…> — awaiting under ## Open recommendations (next free number; never reuse one), and its follow-up as - R<n> check YYYY-MM-DD: <rule> · trip-wire <…> under ## Checks due (the server deletes bullets on close; checks must survive).gooseworks:propose_meta_budget_change / gooseworks:propose_meta_creative_replacement (with reason, evidence as report:<YYYY-MM-DD> or user request, recommendation_ref), or gooseworks:prepare_meta_ad_push (with goose_campaign_id) then push status until it awaits approval. Put each returned change id in its bullet; add each push id to meta_push_ids immediately — the server finds this folder only through that list or goose_campaign_id. Never re-propose an item that already has an id.proposal-and-log.md's one template (numbered moves, no ids) with the real approval links, the pause sentence and the cards; deliver it by its § Delivery, scheduled or not. Stop.Pauses: gooseworks:pause_meta_ad_push only on the marketer's own words in this turn naming the ads (user_instruction), with push_id and meta_ad_ids from ## Meta objects, and recommendation_ref. A bare "yes" → ask for the sentence. Report each ad's state after; anything not confirmed is named.
Links: read change or push status → approved / declined / waiting. Budget changes, swaps and pauses made with recommendation_ref: the server writes the decisions.md block and removes the bullet — never duplicate it. Launches / graduations (no recommendation_ref on prepare_meta_ad_push): when push status shows approved or declined, remove the bullet yourself and append a block with Evidence: push:<id> if none exists. Never say "done" early. A refused budget change (budget moved in Ads Manager): relay, re-read, never retry blind.
Instructions (cards), holds the marketer chose, declines given in chat, and check results (with the verdict word — worked / didn't work / too early / can't tell — in the Outcome line): re-read decisions.md immediately before writing, append a ### YYYY-MM-DD — <title> block (Recommended / Evidence / Decided / Outcome, one line each, date never earlier than the newest block), then delete that recommendation's bullet from state.md (write None. if the section empties).
Save the run record (../run-paid-campaigns/references/run-memory.md), each change id and push id copied into its money-move row.
The rule is ../run-paid-campaigns/references/meta-levers.md § With an ads-platform connector; follow it exactly. Here only the bookkeeping:
9. Staged by steps 1, 2 and 4: each change is an open R<n> bullet and its card. No connector call yet.
10. Apply each one only on its named yes, by that rule.
11. Log after the read-back: re-read decisions.md, append a ### YYYY-MM-DD — <title> block (Recommended and Evidence as usual; Decided: the marketer's words; Outcome: applied <time>, read back <value> — the applied time the outcome check uses, or the failure) and delete the bullet from state.md. In local mode nothing else writes these files, so every block is yours.
Never: approve on the marketer's behalf; pause on your own judgement or a schedule; revert for a kill; break the connector rule; keep a ledger anywhere but these two files (the run record links them and adds only recipes, results and the scorecard).
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
The supplied material contains only placeholder text for the skill’s purpose, inputs, steps, and completion criteria. It does not provide enough information to describe a task or result reliably.
Research supported advertising angles using buyer context, current product facts, customer language, and observed ads. Turn the evidence into static ad, video, or copy ideas, matching video concepts to available recipes. Scores rank ideas for testing; they do not predict results.
Analyze campaign data from Google, Meta, and LinkedIn to find strong and wasteful spend. The skill checks funnel performance and test results, then recommends which campaigns to pause, scale, or test and how to shift budget across channels.
Compare ad spend and signups with downstream outcomes to calculate the cost of each qualified lead. The skill checks tracking coverage, accounts for how long leads need to mature, and groups creatives into actionable categories with confidence notes.
Prepare the intake for a Meta ad campaign by reviewing available brand information first, stating reasonable assumptions, and asking only questions whose answers could affect budget, claims, or legal exposure. It records stable brand facts, a campaign brief, and what was asked versus assumed.
Review ad copy, image text, visuals, and people against platform and brand category rules. It suggests honest fixes and needed disclosures, then flags issues requiring human or legal review, with a separate result for each ad and platform.