Skip to main content
Glama

Make a promo

make_promo

Makes a 15 to 30 s studio-style promo of a web app, a phone app, a new feature or a shop's product from its site or brand kit, in 9:16 first: brand kit (imported from the site if there is none, 2 credits), the app's screens, a recipe, the lines and the checks, then a draft. It needs real screens of the product: public pages of the app itself (app_pages, captured at 2 credits a page) or screenshots and recordings from the person. When it has too few, it renders nothing and says what it needs, with an upload link and a recording link; call it again with promo_id when they're in. The draft is priced per scene (3 credits per started 2 s, nothing for a scene already rendered) plus 10 to put it together: about 43 to 49 credits for a web app, 43 to 49 for a phone app, 46 for a feature launch and 49 for a shop's product for a first 9:16. A phone app takes phone screenshots, taps made from two screenshots (pairs) and store links; a feature launch takes the feature's name and a real screenshot from before it (old_screens); a shop's product takes its product page (product_page), or the person's photos, name and price (product). estimate_only gives the range first and spends nothing. Screenshots the person attached: in Claude, send them from your code sandbox to upload_post_url first (get_promo's next says how), without asking for them again; in ChatGPT pass them in files. For screens not attached yet, add_screens shows an upload box in the chat. Returns the promo as get_promo does.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "make_promo"
  }
}

Input Schema

NameRequiredDescriptionDefault
urlNoThe product's public site, e.g. https://dividendmapper.com
filesNoChatGPT only: images the person attached to the chat. Leave out in other apps
linesNoYour own lines, checked like the writer's: one outcome per feature
modelNoThe line writer's model (owner only)
notesNoWhat to say: the line writer's brief
pairsNoUp to 3 taps or clicks, each from two of the person's screenshots of the same screen (a phone app's tap scene, or a click on a web app): 2 credits each when made, nothing if the control isn't found
sizesNoThe sizes the person wants, for the prices (all four by default)
soundNoMusic, sound effects and voice
lineupNoOne of the recipe's line-ups (list_recipes lists each with what it requires). Left out, the first the material meets; when none is met, the promo says what to send. A Quick line-up (list_recipes marks it) is used only when named, and sets its recipe; the promo stays Quick until a full line-up is named or null is passed, which goes back to the first full line-up the material meets
recipeNoThe recipe: saas (a web app), mobile (a phone app), feature-launch (a new feature of a product) or ecom (a product in an online shop). Left out, the tool picks from what it is given and says which
storesNoA phone app's store pages: the end card shows each store's badge
featureNoThe new feature's name, as the product calls it (feature launch)
productNoThe product as the person gives it: for marketplaces, or a shop capture can't read. Shown as given, marked as theirs. On a promo that has one, the fields left out are kept, even when the name changes: for another product, give each field (null for none)
screensNoUp to 8 screenshots of the app, added to the brand kit
brand_idNoOr a brand kit, by id or name
promo_idNoCarry on with this promo: new screens, or after a needs reply
app_pagesNoUp to 4 public pages that show the product itself working (a public demo, tool or dashboard page), never the home page, product marketing or pricing (those go in site_pages, for a Quick promo), and never a login page; captured at 2 credits a page. For an app behind a login, leave this out and ask the person for screenshots or the recording link
site_pagesNoQuick promos only: up to 3 public pages of the person's own site to make it from, 2 credits a page. Left out, the address itself. Home, product and pricing pages belong here, never in app_pages
old_screensNoScreenshots of the product from before the feature, the same kind of screen as the new ones (feature launch); they are only ever shown as the before
product_pageNoThe product's own page in the shop (ecom), e.g. https://shop.example/products/tote; captured with its photos, price and rating at 2 credits. null takes it off the promo
wait_secondsNoHow long to wait for the promo to settle (default 45)
estimate_onlyNoGive the price range and the screens check (which screens can't be used, by their size), and spend nothing
other_productsNoUp to 3 other product pages from the same shop, for the grid (ecom); 2 credits each

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed7 schema fields changed
    • changedInput schema / properties / app_pages / description
      Previous value: -"Up to 4 public pages that show the product itself working (a public demo, tool or dashboard page), never the home page, product marketing or pricing, and never a login page; captured at 2 credits a page. For an app behind a login, leave this out and ask the person for screenshots or the recording link"New value: +"Up to 4 public pages that show the product itself working (a public demo, tool or dashboard page), never the home page, product marketing or pricing (those go in site_pages, for a Quick promo), and never a login page; captured at 2 credits a page. For an app behind a login, leave this out and ask the person for screenshots or the recording link"
    • addedInput schema / properties / lineup / anyOf
      Added value: +[
      +  {
      +    "maxLength": 40,
      +    "minLength": 1,
      +    "type": "string"
      +  },
      +  {
      +    "type": "null"
      +  }
      +]
    • changedInput schema / properties / lineup / description
      Previous value: -"One of the recipe's line-ups (list_recipes lists each with what it requires). Left out, the first the material meets; when none is met, the promo says what to send"New value: +"One of the recipe's line-ups (list_recipes lists each with what it requires). Left out, the first the material meets; when none is met, the promo says what to send. A Quick line-up (list_recipes marks it) is used only when named, and sets its recipe; the promo stays Quick until a full line-up is named or null is passed, which goes back to the first full line-up the material meets"
    • removedInput schema / properties / lineup / maxLength
      Removed value: -40
    • removedInput schema / properties / lineup / minLength
      Removed value: -1
    • removedInput schema / properties / lineup / type
      Removed value: -"string"
    • addedInput schema / properties / site_pages
      Added value: +{
      +  "description": "Quick promos only: up to 3 public pages of the person's own site to make it from, 2 credits a page. Left out, the address itself. Home, product and pricing pages belong here, never in app_pages",
      +  "items": {
      +    "maxLength": 2048,
      +    "type": "string"
      +  },
      +  "maxItems": 3,
      +  "type": "array"
      +}
  2. Changed1 schema field changed
    • changedInput schema / properties / estimate_only / description
      Previous value: -"Give the price range and spend nothing"New value: +"Give the price range and the screens check (which screens can't be used, by their size), and spend nothing"
  3. First observed

TDQS

A4.4/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Well beyond what readOnlyHint=false and openWorldHint=true convey, the description discloses that the call spends money, gives concrete credit pricing (2 credits per captured page, 3 credits per started 2 s of scene, ~43-49 credits total), states the brand kit is imported from the site when absent, explains the no-render-on-insufficient-screens behavior, and notes the draft returns as get_promo does. That is exactly the cost and side-effect information an agent needs before an irreversible spend.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose and constraints are front-loaded, which is good, but the body is a single dense block with long semicolon-chained clauses mixing pricing, recipe requirements, upload paths and app-specific instructions. Every fact is arguably useful, yet the lack of structure makes it harder to scan than it needs to be.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 23-parameter tool with nested objects and no output schema, the description covers the hard parts: prerequisites, failure mode, pricing, continuation via promo_id, and a pointer that the return matches get_promo. The only soft spot is that the return value is defined only by reference rather than described, though the get_promo pointer is a reasonable substitute.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the schema already carries per-parameter meaning and the baseline is 3. The description still adds cross-parameter semantics the schema cannot: the recipe-to-parameter mapping, the Claude (code sandbox to upload_post_url) versus ChatGPT (files) screenshot path, and that promo_id continues an existing promo. That is real value above the schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The first sentence names a concrete verb and artifact ("Makes a 15 to 30 s studio-style promo"), the inputs it works from ("its site or brand kit"), the four product types, and the default aspect ratio. An agent can immediately tell this is the top-level build-a-promo entry point rather than a scene, caption or render tool.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Guidance is unusually rich: it says per recipe what material is required (phone screenshots and pairs for mobile, feature name plus old_screens for feature launch, product_page or product for ecom), what to do when material is missing (renders nothing, call again with promo_id), and when to use estimate_only. It also routes app_pages vs site_pages by page type. What it does not do is say when to use this versus preview_promo or render_promo, leaving the build/preview/render boundary implicit.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources