Pilot

View saved version →

---
name: anomaly-detection
description: "Scans the ad accounts of a workspace for anomalies through the Pilot MCP server: data gaps, spend spikes or drops, campaigns spending without results, budgets capped or idle, account issues and sudden changes in site traffic, and returns a short alert list ranked by urgency. Use when the user asks if anything looks wrong, for a daily check, or for alerts. Triggers: anomaly detection, anything unusual, daily check, alerts. Déclencheurs : détection d'anomalies, quelque chose d'anormal, contrôle du jour, alertes, vérification quotidienne."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with at least one ad platform attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "pilot_data_freshness pilot_diagnose_wasted_spend pilot_budget_pacing google_list_accounts google_get_campaign_performance google_get_account_spend_today meta_list_accounts meta_get_daily_spend_breakdown meta_check_account_issues google_analytics_run_report"
  pilot-mcp-read-only: "true"
---

# Anomaly detection

Run a quick check of the workspace and return only what looks wrong, ranked by urgency. If nothing is wrong, say so in one line. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace to check.
- **Optional MCP servers:** a team messaging server to post the alerts, a note-taking server to log them. None is needed.
- **Read-only:** never pause a campaign or change a budget, even for an urgent alert. Alerts are for a person to act on.

## Tools

- `pilot_data_freshness`: collection gaps, and accounts that stopped delivering.
- `pilot_diagnose_wasted_spend`: spend without conversions and outlier costs per acquisition.
- `pilot_budget_pacing`: capped and idle budgets.
- `google_list_accounts`: the Google Ads accounts attached to the workspace.
- `google_get_campaign_performance`: recent Google Ads results, day by day when needed.
- `google_get_account_spend_today`: spend today, in micros (divide by 1,000,000).
- `meta_list_accounts`: the Meta ad accounts attached to the workspace.
- `meta_get_daily_spend_breakdown`: daily Meta spend.
- `meta_check_account_issues`: Meta account limits, rejections and restrictions.
- `google_analytics_run_report`: daily sessions and key events, to catch a site or tracking break.

## Steps

1. Call `pilot_data_freshness`. A collection gap or an account that stopped delivering is the first alert.
2. Call `pilot_diagnose_wasted_spend` with `days: 7` and `pilot_budget_pacing` with `days: 7`.
3. Google Ads: `google_list_accounts`, then `google_get_account_spend_today` and `google_get_campaign_performance` on the last 7 days against the 7 before (`start_date`, `end_date`).
4. Meta Ads: `meta_list_accounts`, then `meta_get_daily_spend_breakdown` with `date_preset: "last_14d"` and `meta_check_account_issues`.
5. GA4, if attached: `google_analytics_run_report` with `dimensions: ["date"]` and `metrics: ["sessions", "keyEvents"]` over the last 14 days. A day near zero points to tracking.
6. Flag a change only when it is large against the usual day-to-day variation of the account; mention the threshold you used.

## Output

A list of alerts, most urgent first. Each alert gives what happened, where (platform, account, campaign), since when, the figure against its usual level, and the first thing to check. End with "Nothing else unusual" when the rest is normal.

## Rules

- Rule out a data gap before raising a performance alert.
- Never report a weekend or holiday dip as an anomaly without comparing with the same days before.
- Keep the list short: five alerts at most, the rest in one line.

## Examples

- "Anything unusual in my accounts today?"
- "Fais le contrôle quotidien de l'espace et dis-moi seulement ce qui cloche."
- "Y a-t-il une dépense anormale sur Meta cette semaine ?"

View saved version →

---
name: budget-pacing-check
description: "Checks advertising budgets and spend pacing through the Pilot MCP server: campaigns capped by their budget, budgets left idle, spend so far this month, daily spend trends and the projected effect of a budget change, across Google Ads and Meta Ads. Use when the user asks whether the budget is on track, overspending or underspending. Triggers: budget pacing, are we on budget, monthly spend check. Déclencheurs : rythme du budget, suivi du budget, dépense du mois, budget plafonné, budget sous-utilisé."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with Google Ads or Meta Ads attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "pilot_budget_pacing pilot_cross_platform_overview google_list_accounts google_get_account_budget_summary google_get_account_spend_today google_list_budgets google_simulate_budget_change meta_list_accounts meta_get_daily_spend_breakdown meta_get_minimum_budgets"
  pilot-mcp-read-only: "true"
---

# Budget pacing check

Tell the user whether spend is on track for the month, where budgets cap delivery and where they sit idle. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace concerned.
- **Optional MCP servers:** a team messaging server to send the check, a spreadsheet or file storage server to keep a pacing log. None is needed.
- **Read-only:** never change a budget. Proposed changes are for a person to apply.

## Tools

- `pilot_budget_pacing`: capped, idle and undetermined budgets, with the reason.
- `pilot_cross_platform_overview`: spend so far across platforms.
- `google_list_accounts`: the Google Ads accounts attached to the workspace.
- `google_get_account_budget_summary`: account-level budget and spending limits.
- `google_get_account_spend_today`: spend today, in micros (divide by 1,000,000).
- `google_list_budgets`: campaign budgets and their amounts.
- `google_simulate_budget_change`: projected clicks, cost and conversions at other budget levels for one campaign.
- `meta_list_accounts`: the Meta ad accounts attached to the workspace.
- `meta_get_daily_spend_breakdown`: daily Meta spend.
- `meta_get_minimum_budgets`: Meta's minimum daily budgets.

## Steps

1. Ask for the monthly target if the user has one; otherwise judge pacing against the budgets set in the platforms.
2. Call `pilot_budget_pacing` (default window) to list capped, idle and undetermined budgets. Keep undetermined budgets apart: an unknown budget is not a zero budget.
3. Call `pilot_cross_platform_overview` with the number of days elapsed this month to get spend so far.
4. Google Ads: `google_list_accounts`, then `google_get_account_budget_summary`, `google_list_budgets` and `google_get_account_spend_today`. For a capped campaign that converts well, call `google_simulate_budget_change` to show what more budget would bring.
5. Meta Ads: `meta_list_accounts`, then `meta_get_daily_spend_breakdown` with `date_preset: "this_month"`, and `meta_get_minimum_budgets` when a budget looks too low to deliver.
6. Project month-end spend from the daily average of the days elapsed, and compare it with the target.

## Output

1. **Verdict**: on track, ahead or behind, with spend so far, projected month-end spend and the target.
2. **Capped budgets**: campaigns limited by budget, with their results and, when available, the simulated gain.
3. **Idle budgets**: budgets that are not spent, and why.
4. **Proposed moves**: budget shifts for a person to apply, with the expected effect.

## Rules

- Say which days the projection uses and which currency each amount is in.
- Never sum amounts in different currencies.
- Simulations are Google's estimates: present them as such.

## Examples

- "Are we on track with this month's budget?"
- "Quelles campagnes sont plafonnées par leur budget, et lesquelles dorment ?"
- "Mon budget Meta du mois est-il bien rythmé ?"

View saved version →

---
name: client-meeting-prep
description: "Prepares a client meeting brief through the Pilot MCP server: results for the period across Google Ads, Meta Ads and Google Analytics 4, what worked and what did not, budget status, concentration risk and the recommendations to discuss, written for an account manager to present. Use when the user asks to prepare a client meeting, a client review or talking points. Triggers: prepare client meeting, client review deck notes, talking points. Déclencheurs : préparer un point client, préparer un rendez-vous client, bilan client, points à aborder."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on the client's workspace."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "pilot_data_freshness pilot_cross_platform_overview pilot_budget_pacing pilot_account_concentration google_list_accounts google_get_campaign_performance google_get_recommendations meta_list_accounts meta_get_account_performance_summary google_analytics_run_report"
  pilot-mcp-read-only: "true"
---

# Client meeting prep

Write the brief an account manager reads before a client meeting: the results, the story behind them, and the decisions to ask the client for. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the client's workspace.
- **Optional MCP servers:** a file storage server to save the brief, a note-taking server to keep meeting notes, a team messaging server to share it with the team. None is needed.
- **Read-only:** never change anything in the accounts. Recommendations are for the meeting.

## Tools

- `pilot_data_freshness`: whether the figures are complete.
- `pilot_cross_platform_overview`: consolidated spend and results.
- `pilot_budget_pacing`: budget status.
- `pilot_account_concentration`: how many campaigns carry most of the spend.
- `google_list_accounts`: the Google Ads accounts of the client.
- `google_get_campaign_performance`: Google Ads campaigns over the period.
- `google_get_recommendations`: Google's recommendations to discuss.
- `meta_list_accounts`: the Meta ad accounts of the client.
- `meta_get_account_performance_summary`: Meta account KPIs.
- `google_analytics_run_report`: site traffic and key events by channel.

## Steps

1. Ask for the period if the user did not give one; by default, the last full month.
2. Call `pilot_data_freshness`, then `pilot_cross_platform_overview` over the period.
3. Google Ads: `google_list_accounts`, `google_get_campaign_performance` with `start_date` and `end_date`, and `google_get_recommendations`.
4. Meta Ads: `meta_list_accounts`, then `meta_get_account_performance_summary` with the matching preset (for example `last_month`).
5. GA4, if attached: `google_analytics_run_report` by `sessionDefaultChannelGroup` with `sessions` and `keyEvents`.
6. Call `pilot_budget_pacing` and `pilot_account_concentration` for the risks to raise.
7. Write the brief in plain language for a client, with the numbers that support each point.

## Output

1. **The period in three lines**: spend, results and the trend.
2. **What worked**: two or three points with figures.
3. **What did not**: two or three points with figures and the reason when known.
4. **Budget and risks**: pacing, concentration, anything fragile.
5. **To decide with the client**: the recommendations, each with its expected effect.
6. **Likely questions**: questions the client may ask, with short answers.

## Rules

- Write for a client: no tool names, no jargon without a short explanation.
- State the period and the currency.
- Keep Google's recommendations apart from your own, and present them as suggestions.

## Examples

- "Prepare my meeting with the client tomorrow, last month's results."
- "Prépare le point client de jeudi : résultats d'octobre et recommandations."
- "Quels sont les points à aborder avec ce client sur le mois dernier ?"

View saved version →

---
name: conversion-tracking-check
description: "Checks conversion tracking across platforms through the Pilot MCP server: Google Ads conversion actions and their settings, value rules, Meta pixels and their activity, GA4 key events and the link between GA4 and Google Ads, then compares the volumes each source reports and flags gaps. Use when the user doubts conversion figures, before trusting an optimization, or after a site change. Triggers: conversion tracking audit, pixel check, are my conversions tracked. Déclencheurs : contrôle du suivi des conversions, vérifier le pixel, conversions qui ne remontent pas, audit du tracking."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with Google Ads, Meta Ads or Google Analytics 4 attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "pilot_data_freshness google_list_accounts google_list_conversion_actions_detailed google_list_conversion_value_rules google_get_campaign_performance meta_list_accounts meta_list_pixels meta_get_pixel_stats google_analytics_list_key_events google_analytics_list_google_ads_links google_analytics_run_report"
  pilot-mcp-read-only: "true"
---

# Conversion tracking check

Check that conversions are tracked the way the user thinks, on each platform, and that the platforms roughly agree. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace concerned.
- **Optional MCP servers:** a file storage server to save the checklist, a team messaging server to send it to the developer in charge. None is needed.
- **Read-only:** never create or change a conversion action, a pixel, a key event or a link. Fixes are for a person to apply.

## Tools

- `pilot_data_freshness`: whether Pilot's collected data is current.
- `google_list_accounts`: the Google Ads accounts attached to the workspace.
- `google_list_conversion_actions_detailed`: conversion actions with status, category, counting, windows, value and attribution.
- `google_list_conversion_value_rules`: rules that change conversion values.
- `google_get_campaign_performance`: conversions reported per campaign.
- `meta_list_accounts`: the Meta ad accounts attached to the workspace.
- `meta_list_pixels`: the pixels of the account.
- `meta_get_pixel_stats`: events received by a pixel over time.
- `google_analytics_list_key_events`: the events GA4 counts as conversions.
- `google_analytics_list_google_ads_links`: whether the GA4 property is linked to Google Ads.
- `google_analytics_run_report`: key events by name and by day.

## Steps

1. Call `pilot_data_freshness` to rule out a collection gap.
2. Google Ads: `google_list_accounts`, then `google_list_conversion_actions_detailed`. Check that the primary actions are enabled, counted as expected (every conversion or one per click), with sensible windows, and that the action used for bidding is the right one. Read `google_list_conversion_value_rules`.
3. Call `google_get_campaign_performance` over the last 30 days to see the conversions Google Ads reports.
4. Meta Ads: `meta_list_accounts`, `meta_list_pixels`, then `meta_get_pixel_stats` for each pixel over the last 14 days. A pixel with no event for days is a red flag.
5. GA4: `google_analytics_list_key_events`, `google_analytics_list_google_ads_links`, then `google_analytics_run_report` with `dimensions: ["eventName", "date"]` and `metrics: ["keyEvents"]` over the last 14 days.
6. Compare the volumes. Platforms count differently (attribution, windows, view-through), so look for gaps far beyond those differences, and for sudden breaks on a date.

## Output

1. **Verdict**: tracking looks sound, partly broken, or broken, in one line.
2. **By platform**: what is tracked, how it is counted, and what looks wrong.
3. **Cross-check**: the volumes side by side for the same period, with the expected differences explained.
4. **Fixes**: a checklist for the person in charge, most important first.

## Rules

- Do not expect the platforms to report the same number: explain the usual differences before raising a gap.
- Name the period of every volume.
- Never call a tool that changes tracking settings.

## Examples

- "Are my conversions tracked correctly across Google Ads, Meta and GA4?"
- "Vérifie le suivi des conversions : les achats ne remontent plus depuis lundi."
- "Le pixel Meta reçoit-il encore des événements ?"

View saved version →

---
name: ga4-conversions-and-channels
description: "Analyzes Google Analytics 4 conversions and acquisition channels through the Pilot MCP server: which channels bring sessions and key events, how they change over time, which key events are configured, and where users drop in a funnel. Use when the user asks about GA4, channels, traffic sources, conversions or funnels. Triggers: GA4 analysis, channel performance, traffic sources, conversion funnel. Déclencheurs : analyse GA4, canaux d'acquisition, sources de trafic, conversions Google Analytics, entonnoir de conversion."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with a Google Analytics 4 property attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "google_analytics_get_account_summaries google_analytics_get_metadata google_analytics_list_key_events google_analytics_list_channel_groups google_analytics_check_compatibility google_analytics_run_report google_analytics_run_funnel_report"
  pilot-mcp-read-only: "true"
---

# GA4 conversions and channels

Show which acquisition channels bring traffic and conversions in Google Analytics 4, how that changed, and where users drop. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace where the property is attached.
- **Optional MCP servers:** a file storage server to save the tables, a team messaging server to share them. None is needed.
- **Read-only:** never create or change a key event, audience, channel group or link.

## Tools

- `google_analytics_get_account_summaries`: the properties attached to the workspace.
- `google_analytics_get_metadata`: the dimensions and metrics available on the property.
- `google_analytics_list_key_events`: the events counted as conversions.
- `google_analytics_list_channel_groups`: custom channel groups, if any.
- `google_analytics_check_compatibility`: whether dimensions and metrics can be combined.
- `google_analytics_run_report`: standard reports.
- `google_analytics_run_funnel_report`: funnel reports.

## Steps

1. Call `google_analytics_get_account_summaries`. If several properties are attached, ask which one and pass it as `propertyId`; with one, leave it out.
2. Call `google_analytics_list_key_events` to know what counts as a conversion. If none is configured, say so: conversion figures will be empty.
3. Call `google_analytics_run_report` with two `dateRanges` of equal length (by default the last 28 days and the 28 days before), `dimensions: ["sessionDefaultChannelGroup"]`, `metrics: ["sessions", "engagedSessions", "keyEvents", "totalRevenue"]`.
4. For the channels that moved most, break down by `sessionSource` and `sessionMedium`. When unsure a combination works, call `google_analytics_check_compatibility` first; check names with `google_analytics_get_metadata`.
5. If the user describes a path (landing page, product page, cart, purchase), build it as a funnel with `google_analytics_run_funnel_report`.
6. If custom channel groups exist (`google_analytics_list_channel_groups`), mention which grouping the figures use.

## Output

1. **Summary**: sessions, key events and revenue for both periods.
2. **By channel**: a table with sessions, key events, conversion rate and the change.
3. **What moved**: the sources behind the largest changes.
4. **Funnel**, when asked: step counts and drop-off rates.
5. **Tracking notes**: missing or doubtful key events.

## Rules

- Name the property, the periods and the time zone of the property when it matters.
- A rate with an empty denominator is not available, not zero.
- GA4 attribution differs from the ad platforms': do not expect the numbers to match theirs.

## Examples

- "Which channels brought conversions this quarter?"
- "Compare les canaux GA4 du mois dernier à ceux du mois précédent."
- "Where do users drop between the product page and the purchase?"

View saved version →

---
name: google-ads-account-audit
description: "Audits a Google Ads account in read-only mode through the Pilot MCP server: structure, spend without conversions, search terms, bidding, conversion tracking, ad strength, recommendations and recent changes, then returns a prioritized findings report. Use when the user asks to audit, review or health-check a Google Ads account. Triggers: audit Google Ads, review my Google Ads account, account health check. Déclencheurs : audit Google Ads, auditer un compte Google Ads, bilan de santé du compte, revue de compte Google Ads."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with a Google Ads account attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "google_list_accounts google_get_account_summary google_list_campaigns google_get_campaign_performance google_list_bidding_strategies google_get_search_terms_report google_get_keyword_performance google_list_conversion_actions_detailed google_get_ad_strength google_get_recommendations google_get_optimization_score google_get_change_history pilot_diagnose_wasted_spend pilot_account_concentration pilot_data_freshness"
  pilot-mcp-read-only: "true"
---

# Google Ads account audit

Audit one Google Ads account and hand back a short, prioritized list of findings, each backed by numbers from the account. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace that holds the account.
- **Optional MCP servers:** a file storage server to save the report, a team messaging server to share it, a note-taking server to keep the findings. None is needed to run the audit.
- **Read-only:** never call a tool that creates, updates, pauses or deletes anything, on Pilot or on any other server. Every recommendation is written for a person to apply.

## Tools

- `google_list_accounts`: the accounts attached to the workspace, with their IDs.
- `google_get_account_summary`: account totals for the period.
- `google_list_campaigns`: campaigns, status, channel type and budget.
- `google_get_campaign_performance`: cost, clicks, conversions and value per campaign.
- `google_list_bidding_strategies`: portfolio bidding strategies and their targets.
- `google_get_search_terms_report`: the queries that triggered the ads.
- `google_get_keyword_performance`: keyword metrics.
- `google_list_conversion_actions_detailed`: conversion actions, counting, attribution and status.
- `google_get_ad_strength`: ad strength of responsive search ads.
- `google_get_recommendations`: Google's own recommendations.
- `google_get_optimization_score`: the account optimization score.
- `google_get_change_history`: recent changes, to date any break.
- `pilot_diagnose_wasted_spend`: campaigns that spend without converting, or whose cost per acquisition is far above the account median.
- `pilot_account_concentration`: how many campaigns carry 80% of the spend.
- `pilot_data_freshness`: whether Pilot's collected data is current.

## Steps

1. Call `google_list_accounts`. If several accounts are attached, ask which one to audit, unless the user named it. Use its ID as `customer_id` in every Google Ads call.
2. Call `pilot_data_freshness` with `platform: "google"`. If the data is stale or missing, say so at the top of the report and rely on the live Google Ads tools.
3. Set the period: the last 30 days by default (`date_range: "LAST_30_DAYS"`), or the period the user gave.
4. Read the account: `google_get_account_summary`, `google_list_campaigns`, `google_get_campaign_performance`.
5. Read where money leaks: `pilot_diagnose_wasted_spend` with `platform: "google"`, then `google_get_search_terms_report` and `google_get_keyword_performance` on the campaigns that spend most.
6. Read the setup: `google_list_bidding_strategies`, `google_list_conversion_actions_detailed`, `google_get_ad_strength`.
7. Read the context: `pilot_account_concentration`, `google_get_optimization_score`, `google_get_recommendations`, `google_get_change_history`.
8. Write the report. Rank findings by money at stake. Quote the figure behind each finding, and the period.

## Output

A report with these sections, in the user's language:

1. **Summary**: three lines on the account's state and the top three priorities.
2. **Findings**: a table with the finding, the evidence (figure and period), the estimated stake, and the suggested action.
3. **Tracking**: what the conversion setup counts, and any doubt about it.
4. **Next steps**: the actions to take, in order, for a person to apply in Google Ads.

## Rules

- Never compute a ratio on an empty denominator: write that the value is not available, not zero.
- Do not judge a campaign on fewer conversions than the account usually needs to be stable; say the volume is too low.
- Keep Google's recommendations separate from your own findings, and treat them as suggestions.
- If a tool is refused because an account is not attached, show the list of usable accounts from the message and stop.

## Examples

- "Audit my Google Ads account for the last 30 days."
- "Fais l'audit de mon compte Google Ads, en priorisant ce qui coûte le plus."
- "Quels sont les trois chantiers prioritaires sur le compte Google Ads de la boutique ?"

View saved version →

---
name: meta-ads-creative-audit
description: "Audits Meta Ads creatives and audiences through the Pilot MCP server: best and worst ads, creative fatigue, ad sets with too little volume to judge, audience setup and account issues, then returns what to refresh, keep or stop for a person to apply. Use when the user asks about Meta or Facebook and Instagram ads, creatives, ad fatigue or audiences. Triggers: Meta ads audit, creative fatigue, best performing ads, Facebook ads review. Déclencheurs : audit Meta Ads, fatigue des créas, meilleures publicités, audiences Meta, revue Facebook Ads."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with a Meta ad account attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "meta_list_accounts meta_check_account_issues meta_get_account_insights meta_get_adset_insights meta_get_top_performing_ads meta_get_worst_performing_ads meta_list_adsets meta_list_campaigns meta_list_ad_creatives meta_list_custom_audiences"
  pilot-mcp-read-only: "true"
---

# Meta Ads creative audit

Tell the user which Meta ads and creatives work, which are tiring, which ad sets cannot be judged yet, and what to do next. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace that holds the Meta ad account.
- **Optional MCP servers:** a file storage server to save the audit, a team messaging server to send it to the creative team. None is needed.
- **Read-only:** never pause, edit or duplicate an ad, ad set or campaign. Recommendations are for a person to apply.

## Tools

- `meta_list_accounts`: the Meta ad accounts attached to the workspace.
- `meta_check_account_issues`: limits, rejections and restrictions.
- `meta_get_account_insights`: results of the account, by ad set with `level: "adset"`, week by week with `time_increment: 7`.
- `meta_get_adset_insights`: results of one ad set, for a closer look.
- `meta_get_top_performing_ads`: the best ads on a metric.
- `meta_get_worst_performing_ads`: the weakest ads on a metric.
- `meta_list_adsets`: ad sets of the account, with their status and budget.
- `meta_list_campaigns`: campaigns of the account.
- `meta_list_ad_creatives`: the creatives of the account.
- `meta_list_custom_audiences`: custom audiences and their state.

## Steps

1. Call `meta_list_accounts` and pick the account (ask if several). Use its ID as `account_id`.
2. Call `meta_check_account_issues`: a restricted account explains more than any creative.
3. Call `meta_get_account_insights` with `level: "adset"` and `date_preset: "last_7d"`. Set aside every ad set with too little volume to judge: fewer than about 50 results of its optimization goal over these 7 days, the volume Meta usually needs to leave the learning phase. Do not judge them.
4. Call `meta_get_top_performing_ads` and `meta_get_worst_performing_ads` with `date_preset: "last_30d"`, on the metric that matches the goal (cost per result, ROAS, click-through rate).
5. For the ad sets that remain, call `meta_get_account_insights` with `level: "adset"`, `date_preset: "last_30d"` and `time_increment: 7`, and compare frequency, click-through rate and cost week by week: rising frequency with falling click-through rate is fatigue. Call `meta_get_adset_insights` for an ad set that needs a closer look.
6. Call `meta_list_adsets` and `meta_list_campaigns` to name each ad set and its campaign, with its status and budget.
7. Call `meta_list_ad_creatives` to link the best and worst ads to their formats and messages.
8. Call `meta_list_custom_audiences` to spot audiences too small or not ready.

## Output

1. **Summary**: account state, the creatives that carry the results, the main risk.
2. **Keep**: best ads, with their figures.
3. **Refresh**: tired ads, with the fatigue evidence.
4. **Stop or rework**: weak ads, with their figures.
5. **Not readable yet**: ad sets with too little volume, and when to look again.
6. **Audiences**: issues found.

## Rules

- Never judge an ad set set aside at step 3 for too little volume.
- Name the period, the metric and the currency.
- A ratio with an empty denominator is not available, not zero.

## Examples

- "Which Meta ads are tiring and should be refreshed?"
- "Fais l'audit des créas Meta du dernier mois."
- "Quelles publicités Facebook marchent le mieux sur le coût par achat ?"

View saved version →

---
name: performance-drop-diagnosis
description: "Diagnoses a drop in advertising performance through the Pilot MCP server: checks data freshness first, locates the drop by platform and campaign, then looks for the cause in recent changes, auction pressure, search terms, delivery issues, rejected ads and site traffic. Use when the user reports that conversions, revenue, clicks or ROAS fell, or asks why performance dropped. Triggers: performance drop, why did conversions fall, sudden decline. Déclencheurs : baisse de performance, pourquoi les conversions baissent, chute des résultats, diagnostic d'une baisse."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with Google Ads, Meta Ads or Google Analytics 4 attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "pilot_data_freshness pilot_cross_platform_overview google_list_accounts google_get_campaign_performance google_get_change_history google_get_auction_insights google_get_search_terms_report meta_list_accounts meta_compare_periods_performance meta_get_ad_account_activities meta_check_account_issues meta_get_ad_rejections google_analytics_run_report"
  pilot-mcp-read-only: "true"
---

# Performance drop diagnosis

Find where a drop happened and what most likely caused it, with evidence, before anyone changes a campaign. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace concerned.
- **Optional MCP servers:** a team messaging server to share the diagnosis, a note-taking server to log it. None is needed.
- **Read-only:** never call a tool that creates, updates, pauses or deletes anything. The diagnosis ends with actions for a person to take.

## Tools

- `pilot_data_freshness`: rules out a data gap before diagnosing a drop.
- `pilot_cross_platform_overview`: locates the drop by platform.
- `google_list_accounts`: the Google Ads accounts attached to the workspace.
- `google_get_campaign_performance`: locates the drop by Google Ads campaign.
- `google_get_change_history`: what changed in the account, and when.
- `google_get_auction_insights`: competitors gaining impression share on search and shopping.
- `google_get_search_terms_report`: queries that changed.
- `meta_list_accounts`: the Meta ad accounts attached to the workspace.
- `meta_compare_periods_performance`: Meta results before and after.
- `meta_get_ad_account_activities`: what changed in the Meta account.
- `meta_check_account_issues`: limits, restrictions and delivery problems.
- `meta_get_ad_rejections`: ads stopped by review.
- `google_analytics_run_report`: whether site traffic or key events fell too.

## Steps

1. Call `pilot_data_freshness`. If a platform has a collection gap, the "drop" may be missing data: report it and stop there unless the live tools show the same drop.
2. Call `pilot_cross_platform_overview` for the recent period and the previous one of the same length (for example `days: 7` and `days: 14`). Name the platform where the drop sits.
3. Google Ads: `google_list_accounts`, then `google_get_campaign_performance` on both periods (`start_date` and `end_date`), to find the campaigns behind the drop. Then `google_get_change_history`, `google_get_auction_insights` and `google_get_search_terms_report` on those campaigns.
4. Meta Ads: `meta_list_accounts`, then `meta_compare_periods_performance`, `meta_get_ad_account_activities`, `meta_check_account_issues` and `meta_get_ad_rejections`.
5. If a Google Analytics 4 property is attached, call `google_analytics_run_report` on both periods with `dimensions: ["sessionDefaultChannelGroup"]` and `metrics: ["sessions", "keyEvents"]`. A drop on every channel points to the site or the tracking, not the ads.
6. Rank the candidate causes by how well the dates and the size of the change match.

## Output

1. **Where**: the platform, accounts and campaigns behind the drop, with figures for both periods.
2. **Most likely cause**: one cause, the evidence, and how sure you are.
3. **Other candidates**: causes that remain possible, and the data that would settle them.
4. **What to do**: actions for a person to take, and what to check first.

## Rules

- Rule out a data gap before anything else.
- Compare periods of equal length and say which days they cover.
- Correlation of dates is evidence, not proof: say how sure you are.
- A ratio with an empty denominator is not available, not zero.

## Examples

- "Our conversions dropped this week, why?"
- "Le ROAS de Meta a chuté depuis lundi : qu'est-ce qui a changé ?"
- "Diagnostique la baisse de clics sur Google Ads ce mois-ci."

View saved version →

---
name: period-comparison
description: "Compares two periods across ad platforms and analytics through the Pilot MCP server: spend, results and ratios for Google Ads and Meta Ads, sessions and key events in Google Analytics 4, clicks in Search Console, with the change for each and its main drivers. Use when the user asks to compare two weeks, months, quarters, or this year against last year. Triggers: compare periods, month over month, year over year. Déclencheurs : comparer deux périodes, ce mois contre le mois dernier, évolution sur un an, comparaison de périodes."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with at least one platform attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "pilot_cross_platform_overview google_list_accounts google_get_campaign_performance meta_list_accounts meta_compare_periods_performance google_analytics_run_report google_search_console_get_search_analytics"
  pilot-mcp-read-only: "true"
---

# Period comparison

Put two periods side by side, platform by platform, and explain what drove the difference. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace concerned.
- **Optional MCP servers:** a file storage server or spreadsheet server to keep the table. None is needed.
- **Read-only:** never call a tool that creates, updates, pauses or deletes anything.

## Tools

- `pilot_cross_platform_overview`: consolidated spend and results for a window of days.
- `google_list_accounts`: the Google Ads accounts attached to the workspace.
- `google_get_campaign_performance`: Google Ads campaigns for any start and end dates.
- `meta_list_accounts`: the Meta ad accounts attached to the workspace.
- `meta_compare_periods_performance`: Meta results for two preset periods.
- `google_analytics_run_report`: GA4 figures for two date ranges in one call.
- `google_search_console_get_search_analytics`: organic clicks and impressions for a date range.

## Steps

1. Fix the two periods with exact dates. They should have the same length; if not, say so and compare daily averages.
2. Google Ads: `google_list_accounts`, then `google_get_campaign_performance` with `start_date` and `end_date` for each period.
3. Meta Ads: `meta_list_accounts`, then `meta_compare_periods_performance` with the closest presets (for example `last_7d`, `last_30d`, `last_month`); say when a preset does not match the dates exactly.
4. GA4: `google_analytics_run_report` with both periods in `dateRanges` and the metrics that matter (`sessions`, `keyEvents`, `totalRevenue`).
5. Search Console: `google_search_console_get_search_analytics` for each period, ending two days ago at most.
6. Use `pilot_cross_platform_overview` for a consolidated view when the periods end today.

## Output

1. **Table**: one row per platform and metric, with period A, period B, the change in value and in percent.
2. **Drivers**: the campaigns, channels or queries behind the largest changes.
3. **Caveats**: unequal lengths, presets that differ from the dates, data delays.

## Rules

- Never add amounts in different currencies.
- A change on an empty base is not a percentage: write the values instead.
- Name the exact dates of both periods.

## Examples

- "Compare September with August across Google Ads and Meta."
- "Compare ce trimestre au même trimestre l'an dernier."
- "Qu'est-ce qui explique l'écart entre les deux dernières semaines ?"

View saved version →

---
name: search-console-analysis
description: "Analyzes Google Search Console data through the Pilot MCP server: queries and pages that gain or lose clicks, pages competing for the same query (cannibalization), click-through rate against position, indexing status of key URLs and sitemaps. Use when the user asks about organic search, SEO performance, Search Console queries or pages. Triggers: Search Console analysis, SEO performance, cannibalization, organic clicks. Déclencheurs : analyse Search Console, requêtes SEO, pages qui perdent des clics, cannibalisation, référencement naturel."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with a Search Console site attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "google_search_console_list_sites google_search_console_get_search_analytics google_search_console_get_url_inspection google_search_console_get_sitemaps"
  pilot-mcp-read-only: "true"
---

# Search Console analysis

Read a site's organic search performance and turn it into a short list of pages and queries to work on. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace where the site is attached.
- **Optional MCP servers:** a file storage server to save the tables, a note-taking server to keep the action list. None is needed.
- **Read-only:** never submit a sitemap or request indexing. Actions are for a person to take.

## Tools

- `google_search_console_list_sites`: the sites attached to the workspace.
- `google_search_console_get_search_analytics`: clicks, impressions, click-through rate and position, by query, page, country, device or date.
- `google_search_console_get_url_inspection`: indexing and coverage status of one URL.
- `google_search_console_get_sitemaps`: submitted sitemaps and their state.

## Steps

1. Call `google_search_console_list_sites`. If several sites are attached, ask which one; pass it as `site_url`. With one site, `site_url` can be left out.
2. Set two periods of equal length, by default the last 28 days and the 28 days before. Search Console data lags by about two days: end the recent period two days ago.
3. Call `google_search_console_get_search_analytics` with `dimensions: ["query"]` for both periods; then with `dimensions: ["page"]`.
4. Call it again with `dimensions: ["query", "page"]` on the recent period to find cannibalization: one query whose clicks split across several pages of the site.
5. Find the pages with many impressions and a click-through rate well below what their position suggests: their title and description deserve work.
6. For the top pages that lost clicks, call `google_search_console_get_url_inspection` with `inspection_url`. Call `google_search_console_get_sitemaps` once.

## Output

1. **Summary**: clicks and impressions for both periods, and the change.
2. **Winners and losers**: queries and pages with the largest changes in clicks.
3. **Cannibalization**: queries split across pages, with the page to keep.
4. **Low click-through rate**: pages to rewrite, with position and impressions.
5. **Indexing**: anything wrong found in inspected URLs or sitemaps.
6. **Action list**: ordered for a person to apply.

## Rules

- Use `row_limit` and narrow the dimensions on large sites: responses over 100,000 characters are cut.
- Position is an average: compare it only between periods of the same length.
- Name the period and the search type (`web` by default).

## Examples

- "Which queries lost clicks over the last 28 days?"
- "Analyse la Search Console du site : pages en baisse et cannibalisation."
- "Quelles pages ont beaucoup d'impressions mais peu de clics ?"

View saved version →

---
name: search-terms-analysis
description: "Analyzes Google Ads search terms through the Pilot MCP server: queries that spend without converting, irrelevant intents, terms already excluded, and new keyword candidates, then proposes negative keywords and keyword additions for a person to apply. Use when the user asks about search terms, search queries, negative keywords or wasted search spend. Triggers: search terms report, negative keywords, wasted queries. Déclencheurs : termes de recherche, requêtes Google Ads, mots clés négatifs, exclusions, requêtes qui gaspillent."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with a Google Ads account attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "google_list_accounts google_list_campaigns google_get_search_terms_report google_get_keyword_performance google_list_shared_sets google_list_shared_criteria google_list_customer_negative_criteria pilot_diagnose_wasted_spend"
  pilot-mcp-read-only: "true"
---

# Search terms analysis

Read the queries that triggered Google Ads search ads, find the ones that waste money or miss intent, and propose exclusions and new keywords. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace that holds the account.
- **Optional MCP servers:** a file storage server to save the proposed lists, a team messaging server to send them for review. None is needed for the analysis.
- **Read-only:** never call a tool that creates, updates, pauses or deletes anything. The proposed negative keywords are a list for a person to add in Google Ads.

## Tools

- `google_list_accounts`: the Google Ads accounts attached to the workspace.
- `google_list_campaigns`: campaigns and their channel type, to keep search campaigns.
- `google_get_search_terms_report`: the actual queries, with their metrics.
- `google_get_keyword_performance`: the keywords that matched them.
- `google_list_shared_sets`: shared negative keyword lists.
- `google_list_shared_criteria`: the terms inside a shared list.
- `google_list_customer_negative_criteria`: account-level exclusions.
- `pilot_diagnose_wasted_spend`: campaigns that spend without converting, to focus the analysis.

## Steps

1. Call `google_list_accounts` and pick the account (ask if several are attached). Use its ID as `customer_id`.
2. Call `pilot_diagnose_wasted_spend` with `platform: "google"` and `days: 30` to know which campaigns leak the most.
3. Call `google_list_campaigns` and keep the search campaigns.
4. Call `google_get_search_terms_report` with `date_range: "LAST_30_DAYS"`, first on the campaigns from step 2, then on the account.
5. Read the existing exclusions: `google_list_customer_negative_criteria`, `google_list_shared_sets`, then `google_list_shared_criteria` for each negative list.
6. Call `google_get_keyword_performance` to see which keywords bring the costly queries.
7. Sort the queries into four groups: wasteful (cost and no conversion, or cost per conversion far above the account), off-intent (wrong product, job seekers, free, competitors when unwanted), already excluded but still showing, and promising (converting queries not yet keywords).

## Output

1. **Summary**: spend analyzed, share of spend on queries without conversion, top three actions.
2. **Negative keywords to add**: a table with the term, the proposed match type, the level (campaign, ad group or shared list), the spend and conversions behind it.
3. **Keywords to add**: converting queries that deserve their own keyword, with their metrics.
4. **Watch list**: queries with too little volume to decide.

## Rules

- Never propose excluding a query that converts at an acceptable cost.
- Check every proposed exclusion against existing negatives first.
- Mind close variants: a negative in exact match only blocks that query.
- Show the period and the currency of every amount.

## Examples

- "Which search terms wasted money last month?"
- "Analyse les termes de recherche de mes campagnes Search et propose des mots clés négatifs."
- "Quelles requêtes convertissent mais ne sont pas encore des mots clés ?"

View saved version →

---
name: weekly-performance-report
description: "Builds a weekly multi-platform advertising report through the Pilot MCP server: consolidated spend and results across Google Ads and Meta Ads, week-over-week changes, budget pacing and the traffic seen in Google Analytics 4. Use when the user asks for a weekly report, a weekly recap or last week's performance. Triggers: weekly report, weekly recap, how did last week go. Déclencheurs : rapport hebdomadaire, bilan de la semaine, récap hebdo, point de la semaine dernière."
compatibility: "Requires the Pilot MCP server (https://app.pilot.braike.com/mcp) connected in the client and authorized on a workspace with at least one ad platform attached."
metadata:
  pilot-mcp-server: "https://app.pilot.braike.com/mcp"
  pilot-mcp-tools: "pilot_data_freshness pilot_cross_platform_overview pilot_budget_pacing google_list_accounts google_get_campaign_performance meta_list_accounts meta_get_account_performance_summary meta_compare_periods_performance google_analytics_run_report"
  pilot-mcp-read-only: "true"
---

# Weekly performance report

Produce a one-page weekly report across the ad platforms of the workspace: what was spent, what it brought, what changed against the previous week, and what to watch. Answer in the user's language.

## Requirements

- **Required MCP server:** Pilot MCP (`https://app.pilot.braike.com/mcp`), authorized on the workspace to report on.
- **Optional MCP servers:** a file storage server to save the report, a team messaging server to post it to a channel, a note-taking server to archive it. None is needed to build the report.
- **Read-only:** never call a tool that creates, updates, pauses or deletes anything, on Pilot or on any other server.

## Tools

- `pilot_data_freshness`: whether the collected data covers the week.
- `pilot_cross_platform_overview`: spend and results consolidated across platforms, ratios computed on totals.
- `pilot_budget_pacing`: campaigns capped by their budget, and budgets left idle.
- `google_list_accounts`: the Google Ads accounts attached to the workspace.
- `google_get_campaign_performance`: Google Ads campaigns over the week.
- `meta_list_accounts`: the Meta ad accounts attached to the workspace.
- `meta_get_account_performance_summary`: Meta account KPIs over the week.
- `meta_compare_periods_performance`: Meta results this week against the previous week.
- `google_analytics_run_report`: sessions and key events by channel in Google Analytics 4.

## Steps

1. Call `pilot_data_freshness`. If a platform is stale, say so at the top and treat its figures as partial.
2. Call `pilot_cross_platform_overview` with `days: 7`, then with `days: 14`; derive the previous week as the difference. If the tool refuses a money total because currencies differ, report each currency separately.
3. Google Ads: call `google_list_accounts`, then `google_get_campaign_performance` with `date_range: "LAST_7_DAYS"` for each attached account. Keep the five campaigns with the most spend and the largest changes.
4. Meta Ads: call `meta_list_accounts`, then `meta_get_account_performance_summary` with `date_preset: "last_7d"` and `meta_compare_periods_performance` with `period1: "last_7d"` and `period2: "last_14d"`. The second period contains the first: the previous week is the difference.
5. Call `pilot_budget_pacing` with `days: 7` to flag capped and idle budgets.
6. If a Google Analytics 4 property is attached, call `google_analytics_run_report` with `dateRanges` for the last 7 days and the 7 days before, `dimensions: ["sessionDefaultChannelGroup"]` and `metrics: ["sessions", "keyEvents"]`.
7. Write the report.

## Output

1. **Headline**: total spend, total conversions or value, and the change against last week, in one sentence.
2. **By platform**: a table with spend, conversions, cost per conversion and the week-over-week change.
3. **What moved**: the three largest changes, with their likely cause when the data shows it.
4. **Budget**: capped and idle budgets.
5. **To watch next week**: two or three points.

## Rules

- State the dates covered and the currency of every amount.
- Never add amounts in different currencies.
- A ratio with an empty denominator is not available, not zero.
- Skip a platform the workspace has not attached, without treating it as an error.

## Examples

- "Give me the weekly report for this workspace."
- "Fais-moi le rapport hebdomadaire Google Ads et Meta, avec les variations."
- "How did last week go compared with the week before?"