Build a video dashboard in Claude

Builds the Get MCP Ads Video template as a Claude dashboard for Meta, Google Ads (Video and Demand Gen), TikTok, Pinterest and Snapchat: per video creative with its thumbnail, the hook (each platform's own 2 or 3-second view over impressions), the hold (25 to 100% points, ThruPlays and the like, watch time) and what the same ads cost and returned (CPM, CTR, CPA, ROAS, cost per hook view), against the comparison period the reader picks, with the video share of the account's spend. Ads mixing videos and images are shown apart.

Recipe

Intent A live page showing whether the video ads hold attention and whether that attention turns into results, so the team can cut, re-cut or fund each video.

Steps: 3
Tools: 1
Rules: 6
Failure modes: 3

skills/video-dashboard.md

When to use it

Use it when

Someone asks whether their videos hook people, where viewers drop off, which videos to cut or re-edit, or for a hook rate and hold rate report they can keep in Claude.
Before cutting a video that loses people in its first seconds, re-cutting a strong hook with a weak hold, or moving budget to the videos whose attention converts, and to check afterwards what changed.

Do not use it when

Judging what a video shows: the dashboard reads delivery figures, not the video itself; look at the creative with the creative tools.
Image ads and catalog ads: they are counted in the account's spend, not as videos.
Hosts without Claude's Dashboard artifact type: answer in the conversation with the matching reports instead.

What it needs first

Optional Meta Ads, Google Ads, TikTok Ads, Pinterest Ads, Snapchat Ads

No single source is required: this one reads across whatever you have connected.

Preconditions

The organisation is on Team or Enterprise: ready-made dashboards are not offered on Free or Pro.
The account is among the chosen accounts, confirmed via list_accounts.

Evidence every conclusion must carry

Every figure carries its account, metric and window.
Anything below a readable sample size is reported as unjudged, not as zero.

Inputs
Input	Required	What it is
`window`		Analysis window.Default: last 30 days.

Procedure

In order. Every tool named is one the gateway ships, and links to its reference.

Steps

Resolve the client's workspace and its ad accounts, one per platform (Meta, Google Ads, TikTok, Pinterest, Snapchat), with list_accounts.
Fetch the video template with every account (accounts, one per platform, each with its currency).get_dashboard_template
Create a Dashboard artifact, write every document exactly as given in one batch, read its check, hand over with the lines given.

Decision rules

The observable condition, what it lets you conclude, and what takes the conclusion back.

When The template is refused with PLAN_REQUIREDConclude Say the dashboards come with Team and Enterprise and that only the account owner changes the plan; answer the question in the conversation meanwhile.
When The person compares hook rates across platformsConclude Say each platform measures its own view (Meta 3 seconds, TikTok, Pinterest and Snapchat 2 seconds with their own conditions, Google Ads no short view at all): compare videos within a platform, never one platform's rate with another's.
When A video hooks well and holds poorlyConclude Name it as a re-cut candidate: keep its opening, shorten or restructure what follows; a weak hook with a fair hold is a new opening, a weak hook and hold a cut. Weigh spend and conversions before recommending, and say these are observed rates, not a test.
When A row shows as mixedConclude It is a carousel, flexible ad or Story mixing videos and images: its rates over impressions are left blank because image cards share its impressions; judge it on its own CPA and ROAS, never against single videos.
When The client advertises on Google Ads with Performance MaxConclude Say Performance Max videos are not in the page: Google reports no quartiles or watch time per asset there.
When A reader sees a source fail to loadConclude Their own Get MCP Ads access decides: check that their connection covers this client's workspace and that their plan includes dashboards before changing the dashboard.

Where the agent stops

The readout changes nothing. Any change it motivates goes through Safe Writes: preview, human confirmation, then apply.

How it goes wrong quietly

The cases where the analysis is wrong and still looks right. Read them before trusting a number.

Editing the template's page or queries by hand loses the provenance of each figure: ask for changes by name instead.
Reading a TikTok 2-second rate against a Meta 3-second rate as the same measure, or adding views, conversions or watch time across platforms.
Calling a video new when the comparison was not read whole: the dashboard says so in its notes, and only then is a video without comparison figures new.