Reference for every public CloudX MCP tool, including parameters, supported filters, and response fields.

Tool Reference

CloudX MCP currently exposes 24 public tools: eight reporting tools, four auction-inspection tools, five configuration tools, six Task tools, and one documentation search tool. The authenticated user is always scoped to their CloudX account.

Tool Summary

Reporting Tool Parameters

GetReportDashboard

Use this tool for top-line reporting across a time range.

Best For

• total revenue and fill rate checks
• daily trend analysis
• quick health checks by country or platform

Parameters

Response Shape

• summary
• chart_data

summary includes fields such as total_requests, total_impressions, revenue, fill_rate, ecpm, total_clicks, ctr, and total_users.

GetReportBreakdown

Use this tool to choose the grouping dimensions and metrics for a report instead of using a fixed dashboard shape.

Parameters

Response Shape

• dimensions: requested dimension names
• metrics: requested metric names
• rows: objects containing dimensions and metrics maps
• row_count

For dimension and metric definitions, see cloudx report breakdown.

GetReportBidders

Use this tool when you want to compare demand partners.

Best For

• finding the highest-revenue bidder
• checking bid rate and win rate by partner
• comparing supply quality across countries or platforms

Parameters

Response Shape

• bidders

Each bidder row can include name, requests, bids, bid_rate, impressions, win_rate, revenue, and ecpm.

GetReportApps

Use this tool to compare app-level performance within an account.

Best For

• ranking apps by revenue
• comparing iOS and Android portfolio performance
• spotting low fill-rate apps

Parameters

Response Shape

• apps

Each app row can include app_id, name, platform, impressions, fill_rate, revenue, and ecpm.

GetReportAdUnits

Use this tool for placement-level investigation.

Best For

• comparing banners vs rewarded vs interstitial
• isolating a single app bundle
• finding weak or high-value ad placements

Parameters

Response Shape

• ad_units

Each row can include ad_unit_id, name, app_name, app_bundle, type, impressions, fill_rate, revenue, and ecpm.

GetReportExport

Use this tool when you need tabular data rather than a summarized report.

Best For

• passing data into spreadsheets
• validating downstream analytics
• handing structured rows to another tool or agent

Parameters

Response Shape

• columns
• rows
• row_count

GetReportABTest

Use this tool to compare an A/B test’s control and test variants over a time range.

Parameters

The requested range cannot exceed 31 days.

Response Shape

• control and test, each with cumulative metrics and daily rows
• statistical fields when available: lift, p_value, chi_squared_p, ci_low, ci_high, n_control, n_test, and required_n
• status, such as in_progress, directional, achieved, or invalidated

GetReportTotalRevenue

Use this tool to read the Total Publisher Revenue report shown on the Total Revenue dashboard. It combines CloudX with the mediation platforms you connected in Data Sources.

Best For

• total publisher revenue and CloudX share of wallet
• comparing mediation platforms by app, format, or country
• daily revenue trends that include traffic outside CloudX

Parameters

Mediator reports sync daily. Before 12:00 UTC, data is complete through the end of the UTC day before yesterday; from 12:00 UTC, through the end of yesterday.

Response Shape

• totals: one value per requested measure
• series: daily points
• dimensionTables: one table per requested dimension
• measures, dimensions, availableMeasures, availableDimensions
• startTime, endTime, requestedEndTime, dataCompleteThrough
• reportLink: a link to the same view in the dashboard

Auction Inspection Tools

Auction time-range tools accept Unix timestamps in seconds and a maximum range of 31 days. Their common optional filters are test_mode, app_bundle, ad_unit_id, country, and device_os. Row limits default to 50 and accept values from 1 to 500.

GetAuctionList

Use this tool to find recent auctions before drilling into one auction.

Parameters

has_ilrd and has_external_ilrd are mutually exclusive. mediator can only be used with has_external_ilrd: true.

The response contains an auctions array with IDs, timestamps, app and ad-unit context, winner details, bid counts, duration, A/B test context, and ILRD availability.

GetAuctionShow

Use this tool after GetAuctionList when you know the auction ID.

Parameters

The response includes auction_id and auctions. Requested related data appears in rounds, bids, ilrd, and external_ilrd.

GetAuctionRounds

Use this tool either to inspect rounds for one auction_id or to aggregate round behavior over a time range.

Parameters

The response contains detailed rounds, aggregate metrics, or both.

GetAuctionBids

Use this tool to inspect bid and non-bid rows across a time range.

Parameters

The bids response array includes bidder, status, price, winner, latency, floor, rejection, and auction context.

Configuration Tools

GetConfigShow

Fetch the live config by omitting all parameters, or select a draft or published version.

The response includes id, kind, version and author metadata, parsed data, and yaml.

GetConfigValidate

Validate the live config by omitting all parameters, or select a config with id or version.

The response includes valid, error_count, warning_count, issues, source, and yaml. Each issue can include its severity, code, message, path, line, and column.

GetConfigHistory

List recent configuration rows, newest first. This returns at most 200 rows and is not an exhaustive account history.

Each entries row includes the config ID, kind, version, author, creation time, description, and publish-time diff counts.

PostConfigEdit

Apply typed changes to apps, ad units, ad unit groups, account bidders, network mappings, lists, tags, test devices, line items, or A/B tests.

Request Shape

Supported actions are:

• apps: create_app, update_app, delete_app
• ad units: create_ad_unit, update_ad_unit, delete_ad_unit
• network and bidder settings: upsert_network_mapping, delete_network_mapping, upsert_account_bidder
• ad unit groups: create_ad_unit_group, update_ad_unit_group, delete_ad_unit_group, remove_ad_unit_from_group
• lists and tags: create_list, update_list, delete_list, create_tag, update_tag, delete_tag
• test devices: upsert_test_device, delete_test_device
• A/B tests: create_ab_test, update_ab_test, start_ab_test, end_ab_test, promote_ab_test, delete_ab_test
• line items: create_line_item, update_line_item, delete_line_item

Ordinary edits and create_ab_test return a draft. The A/B test update, start, end, promote, and delete actions publish immediately and must be the only operation in the request. Set price floors on line items; ad-unit create and update operations do not accept bidfloor. Line-item bidfloor values are decimal USD CPM.

PostConfigPublish

Publish a draft as the live configuration. The same server-side checks used by the CloudX app run before the draft becomes live.

The response includes config_id, version_number, and validation. A blocking validation failure returns an MCP error and does not publish the draft.

Task Tools

Use these tools to manage recurring Tasks from an MCP client. A Task runs its prompt on a five-field cron schedule in an IANA timezone. Tasks are private by default; set access_mode to workspace to share one. A private Task is visible only to its creator.

For creation, name defaults to a name derived from the prompt, timezone defaults to UTC, enabled defaults to true, and access_mode defaults to private. The schedule field accepts a five-field cron expression. Updates preserve omitted fields. Only the Task creator can change a workspace Task to private.

Task objects include id, name, prompt, cron_expression, timezone, schedule_description, enabled, access_mode, next_run_at, and timestamps. The list response wraps Task objects in scheduled_agents.

{
  "name": "CreateScheduledAgent",
  "arguments": {
    "body": {
      "prompt": "Review weekly revenue and flag material changes",
      "schedule": "0 9 * * 1",
      "timezone": "Europe/Madrid"
    }
  }
}

RunScheduledAgent requires an enabled Task with no active run. DeleteScheduledAgent rejects a Task with an active run.

SearchDocs

Use this tool to search CloudX documentation from an MCP client before answering setup, configuration, SDK, CLI, dashboard, network, or reporting questions.

Best For

• finding the right documentation page for a CloudX workflow
• grounding an answer in current docs before explaining a feature
• retrieving relevant setup or troubleshooting snippets without leaving the AI client

Parameters

Tool Call Shape

{
  "name": "SearchDocs",
  "arguments": {
    "body": {
      "query": "report filters"
    }
  }
}

Response Shape

• results

Each result includes:

• content: matching documentation snippet
• path: documentation path such as /en/cli/reporting
• url: absolute docs URL
• metadata: page metadata such as title, when available

Error Behavior

These tools return MCP error results when the request cannot be fulfilled. Common causes include:

• missing start_time or end_time
• invalid enum values such as device_os: "ios" instead of iOS
• a time range longer than 31 days
• a missing auction or config ID
• insufficient permission for the requested tool
• config edit or publish validation failures

For SearchDocs, a blank or missing query returns query is required. If documentation search is temporarily unavailable, the tool returns a docs search error.

Related

• Installation
• Examples