run_page_audit optimizes an existing page of the brand for one tracked prompt. Mentionable reads the page as it is live, extracts the searches ChatGPT runs on the prompt (fan-out queries), then rewrites the page so the AI engines can cite it: same voice, same offers, plus what it lacks. The result is the complete page ready to paste, 5 to 7 ranked changes explaining why, and the passages to check on the live page.

The tool returns an auditId right away. The work runs asynchronously (typically 1 to 2 minutes): poll get_page_audit until status is COMPLETED.

Requires the member role minimum. The customer role is rejected. The MCP is available on the Pro and Agency plans.

When to use

Use it when a page of the brand is close to being cited: the engines open it on a prompt but quote another source. list_page_audits shows what was already optimized; the app lists these pages under Pages to optimize. Pass a promptId when you know which question the page should win; without it, the tool picks the prompt where the engines open the page most without citing it.

Billing

Every plan is credit-based. run_page_audit charges 100 credits, debited from the workspace's monthly credit pool, with overage capped by the workspace's monthly overage cap. The fan-out extraction it runs when the prompt has none under 30 days old is included: no extra credit.

Credits are checked before anything starts, and the page is read before the charge: an unreachable page costs nothing.

Input

Field	Type	Default	Description
`projectId`	string (CUID)	required	Project scope.
`pageUrl`	string (URL)	required	Full `https://` URL of a publicly accessible page of the brand.
`promptId`	string (CUID)	the prompt where the engines open the page without citing it	Tracked prompt to optimize the page for.

Response

Success: optimization queued

{
  "success": true,
  "auditId": "clx_page_42",
  "status": "PENDING",
  "promptId": "clx_prompt_7",
  "creditsCharged": 100,
  "remainingCredits": 2900,
  "message": "Optimization queued. Poll get_page_audit with this auditId until status is COMPLETED (typically 1 to 2 minutes)."
}

remainingCredits is null for an uncapped workspace.

Errors

Every error has success: false and a machine code in error:

`error`	Meaning
`insufficient_credits`	Out of credits or capped. Carries `creditCost`, `remainingCredits`, `upgradeUrl`. Nothing started.
`prompt_required`	No `promptId` given and no tracked prompt has met the page yet. Pass one (see `list_prompts`).
`prompt_not_found`	The `promptId` is not an active prompt of this project.
`fetch_failed`	The page could not be read (private, redirecting, blocking crawlers, no text). No credit charged.
`project_not_found`	The project does not exist.

{
  "success": false,
  "error": "insufficient_credits",
  "creditCost": 100,
  "remainingCredits": 40,
  "upgradeUrl": "/billing"
}

Tips and patterns

Start from the gaps. Pages the engines open without citing are the cheapest citations to win: optimize those first.
De-duplicate. Call list_page_audits first and skip a page already optimized for the same prompt.
Poll, don't block. Call get_page_audit every 15 to 20 seconds. The work is bound by the extraction and two writing calls, not by your polling.
Close the loop. Once the user has updated the page, the app's I applied the changes marks it and tracks the citations from the next scan on; get_page_audit returns appliedAt.

Related tools

get_page_audit: fetch the result.
list_page_audits: see what was already optimized.
list_prompts: find the promptId to target.
list_fan_outs: the searches behind a prompt.