Get recommendations (adsap_get_recommendations)
Platform: Meta and Google Ads · Read or write: Read · Plan: Early Access · Consumes a task: No · Preview supported: No
What it does
Returns the alerts and findings that Adsap’s daily and weekly checks computed from your own account data. Alerts flag things that broke or changed and need attention now, such as an account that stopped spending, spend with no conversions, or tracking that went silent. Findings are clear actions worth taking, such as starved ads, losing ads, weak launches, creative fatigue, wasted Google search terms and keyword promotion candidates. Each item comes with evidence numbers, money at risk, and the latest weekly account summary.
When to use it
- Start here when you ask what needs your attention, what to fix, or whether anything is broken.
- It returns two separate lists.
alertsmeans something broke or changed and needs attention now.findingsmeans a clear action worth taking. The assistant leads with the alerts. - It reads results that Adsap’s daily and weekly checks already stored. It makes no call to Meta or Google Ads.
- Leave
account_idout for every account you can reach in the workspace, on both platforms. Pass a Meta ID withact_, or a 10-digit Google Ads customer ID, for one account. limitapplies to each list on its own, 50 by default and 100 at most. Setstatustoresolvedfor past items, andinclude_dismissedtotrueto add dismissed ones.
Worked example
You ask
“Is anything broken on act_1234567890, and what should I fix first?”
The assistant calls adsap_get_recommendations
{
"account_id": "act_1234567890"
}Adsap returns
The field names are the ones Adsap returns. The values are examples.
{
"findings": [
{
"check_id": "losing_ad",
"kind": "diagnostic",
"entity_name": "Spring sale, video 1",
"severity": "high",
"money_at_risk": 412.6,
"currency": "USD",
"evidence": {
"variant": "high_cpa",
"spend": 412.6,
"conversions": 3,
"cpa": 137.53,
"ref_cpa": 48.2,
"ref_scope": "ad_set"
}
}
],
"alerts": [
{
"check_id": "zero_spend_yesterday",
"kind": "alert",
"entity_type": "account",
"entity_id": "act_1234567890",
"severity": "high",
"money_at_risk": null,
"evidence": {
"as_of": "2026-09-17",
"days_with_spend_prior": 7,
"avg_daily_spend_prior": 186.4,
"last_spend_date": "2026-09-16"
}
}
],
"weekly_summaries": []
}How to read it
Read alerts first. Here the account spent on each of the previous 7 days, about 186.40 a day, and spent nothing on 2026-09-17. The assistant explains each item from its evidence numbers and suggests what to check. Alerts are sorted by severity, then by money_at_risk. An alert often has no money figure because nothing is being spent.
findings are sorted by money_at_risk. This ad spent 412.60 for 3 conversions, a cost of 137.53 each, against 48.20 for its ad set. The evidence fields change with each check_id. Every item in both lists has the same fields. This example trims some of them: platform, ad_account_id, entity_type, entity_id, entity_name, currency, status, first_seen_date, last_seen_date, resolved_date, days_since_first_seen and dismissed. weekly_summaries holds the latest weekly note per account when one exists, with a summary_source of ai or deterministic.
Parameters
Descriptions are shortened from the tool schema. Your assistant sees the full text.
| Name | Type | Required | Description |
|---|---|---|---|
account_id | string | No | Meta act_… or bare 10-digit Google Ads customer id. Omit for all reachable accounts in the workspace. |
platform | one of meta, google | No | |
status | one of active, resolved | No | |
include_dismissed | boolean | No | |
limit | number | No |
Example prompts
- “What needs my attention across all my accounts this week?”
- “Is anything broken on my Meta account?”
- “What should I fix on my Google Ads account?”
- “Show me the recommendations I dismissed last month.”
Notes
Leave the account out to get a whole-workspace view across both platforms. The assistant labels an AI-written weekly summary “AI Generated” when it relays one.
Related tools
adsap_list_automation_rules: List the automation rules on one ad account, and optionally what they have done.adsap_manage_automation_rule: Create, change, switch on/off or delete an automation rule for one ad account.
Guide
Read Recommendations for the workflow around this tool, and the Tool reference for every tool.
The Meta Ads MCP server and Google Ads MCP server pages show how to connect each platform to Claude, ChatGPT or Perplexity.