Skip to main content
Glama
hermoso-ai

Hermoso

Official

Plan an ad concept

plan_ad

Convert a brand and product brief into a complete ad concept: copy variants, an image prompt or video storyboard, and the model IDs to render.

Instructions

Creative director: turn a brand + product/brief into a finished ad CONCEPT — copy variants (headline/primary/cta) plus an image_concept.prompt OR a video_storyboard, with the resolved recipe + the model ids to render with. Renders nothing; chain its output into generate_image / generate_video. THE USER’S EXPLICIT LENGTH IS SOVEREIGN: when they name a duration ("a 30 second ad", "make it 45s"), pass it as durationSeconds — the board is then AUTHORED to that length (its scenes sum to it) and render_ad renders it as one clip or stitched acts accordingly. Leaving it out lets the planner pick its own default, which is how an explicit ask silently becomes a 15s spot. Spends credits.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
hookNoforce the VISUAL scroll-stop mechanic the opening beat is built on — a hook id from list_hooks (e.g. "direct_callout", "mid_problem", "macro_asmr"). Omit to let the planner pick. A hook that cannot be delivered in this brief is DROPPED with the reason rather than rendered wrongly — an on-screen-text hook on an authentic/UGC ad is the one that bites, because that register carries zero on-screen text.
brandNobrand name, or a brand profile object {name,domain,category,palette,products,…}. OMIT to use the workspace’s SAVED brand + memory automatically (see get_brand); use draft_brand to onboard a new one
draftNoONLY after a video refusal that offered a light draft: the {model, durationSeconds} it named. The plan is then authored to that length and priced on that model. Never invent one — a video the account cannot cover is refused BEFORE planning with the three options (image / add credits / this draft when one fits), and the user chooses.
formatNo'image', 'video', or 'auto' when unspecified
recipeNoa recipe id from hermoso_capabilities to force an archetype
productYeswhat to advertise + any angle/offer the user specified
settingNoforce the WHERE — a setting id from list_hooks (e.g. "kitchen", "gym", or a surreal one like "volcano_rim" / "airplane_wing", which are played 100% straight and never acknowledged). Omit for a neutral setting.
languageNooutput language for the ad copy (e.g. Spanish) — default English
referenceNoa reference to clone: an ad-library link (Facebook Ad Library, LinkedIn Ad Library, Google Ads Transparency — its real copy/advertiser are fetched) OR a VIDEO link — a TikTok, Instagram Reel, Facebook video, X post, YouTube Short/video or a direct video file — which is WATCHED first (frames + voiceover/on-screen-text transcript) so the concept keeps its hook, structure and pacing. To remake one video for this brand at its own length, clone_video is the direct tool
durationSecondsNoVIDEO ONLY — the total spot length the user explicitly asked for, in seconds, copied verbatim (30 for "a 30 second ad"). The planner authors the storyboard TO it: the scenes’ seconds sum to it and the script is word-budgeted for it. Supported range 4–180; anything outside is CLAMPED to it (the reply says so). A length that fits ONE clip of the render model renders as a single continuous pass; anything longer is STITCHED from acts filled to that model’s clip maximum with the remainder last (on a 15s-clip model, 40 -> 15+15+10 and 17 -> 13+4) — never time-compressed. The maximum is the model’s own: 15s on most, 30s on the longest-clip model, so a 30s ad can be one unbroken take rather than two acts. Omit when the user named no length; do NOT pass a guess, an omitted value keeps the recipe-aware default.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed2 schema fields changedv0.1.272
    • addedInput schema / properties / draft
      Added value: +{
      +  "description": "ONLY after a video refusal that offered a light draft: the {model, durationSeconds} it named. The plan is then authored to that length and priced on that model. Never invent one — a video the account cannot cover is refused BEFORE planning with the three options (image / add credits / this draft when one fits), and the user chooses.",
      +  "properties": {
      +    "durationSeconds": {
      +      "type": "number"
      +    },
      +    "model": {
      +      "type": "string"
      +    }
      +  },
      +  "required": [
      +    "model",
      +    "durationSeconds"
      +  ],
      +  "type": "object"
      +}
    • changedInput schema / properties / durationSeconds / description
      Previous value: -"VIDEO ONLY — the total spot length the user explicitly asked for, in seconds, copied verbatim (30 for \"a 30 second ad\"). The planner authors the storyboard TO it: the scenes’ seconds sum to it and the script is word-budgeted for it. Supported range 4–180; anything outside is CLAMPED to it (the reply says so). A length that fits ONE clip of the render model renders as a single continuous pass; anything longer is STITCHED from acts filled to that model’s clip maximum with the remainder last (on a 15s-clip model, 40 → 15+15+10 and 17 → 13+4) — never time-compressed. The maximum is the model’s own: 15s on most, 30s on the longest-clip model, so a 30s ad can be one unbroken take rather than two acts. Omit when the user named no length; do NOT pass a guess, an omitted value keeps the recipe-aware default."New value: +"VIDEO ONLY — the total spot length the user explicitly asked for, in seconds, copied verbatim (30 for \"a 30 second ad\"). The planner authors the storyboard TO it: the scenes’ seconds sum to it and the script is word-budgeted for it. Supported range 4–180; anything outside is CLAMPED to it (the reply says so). A length that fits ONE clip of the render model renders as a single continuous pass; anything longer is STITCHED from acts filled to that model’s clip maximum with the remainder last (on a 15s-clip model, 40 -> 15+15+10 and 17 -> 13+4) — never time-compressed. The maximum is the model’s own: 15s on most, 30s on the longest-clip model, so a 30s ad can be one unbroken take rather than two acts. Omit when the user named no length; do NOT pass a guess, an omitted value keeps the recipe-aware default."
  2. Changed1 schema field changedv0.1.227
    • changedInput schema / properties / reference / description
      Previous value: -"a reference to remix: an ad-library link (Facebook Ad Library, LinkedIn Ad Library, Google Ads Transparency — its real copy/advertiser are fetched) OR a VIDEO link — a TikTok, Instagram Reel, Facebook video, X post, YouTube Short/video or a direct video file — which is WATCHED first (frames + voiceover/on-screen-text transcript) so the concept keeps its hook, structure and pacing. To remake one video for this brand at its own length, clone_video is the direct tool"New value: +"a reference to clone: an ad-library link (Facebook Ad Library, LinkedIn Ad Library, Google Ads Transparency — its real copy/advertiser are fetched) OR a VIDEO link — a TikTok, Instagram Reel, Facebook video, X post, YouTube Short/video or a direct video file — which is WATCHED first (frames + voiceover/on-screen-text transcript) so the concept keeps its hook, structure and pacing. To remake one video for this brand at its own length, clone_video is the direct tool"
  3. Changed1 schema field changedv0.1.225
    • changedInput schema / properties / reference / description
      Previous value: -"a reference ad URL to remix the angle from — Facebook Ad Library, LinkedIn Ad Library or Google Ads Transparency links (the real ad’s copy/advertiser are fetched and fed into the concept)"New value: +"a reference to remix: an ad-library link (Facebook Ad Library, LinkedIn Ad Library, Google Ads Transparency — its real copy/advertiser are fetched) OR a VIDEO link — a TikTok, Instagram Reel, Facebook video, X post, YouTube Short/video or a direct video file — which is WATCHED first (frames + voiceover/on-screen-text transcript) so the concept keeps its hook, structure and pacing. To remake one video for this brand at its own length, clone_video is the direct tool"
  4. Changed4 schema fields changedv0.1.161
    • addedInput schema / properties / durationSeconds
      Added value: +{
      +  "description": "VIDEO ONLY — the total spot length the user explicitly asked for, in seconds, copied verbatim (30 for \"a 30 second ad\"). The planner authors the storyboard TO it: the scenes’ seconds sum to it and the script is word-budgeted for it. Supported range 4–180; anything outside is CLAMPED to it (the reply says so). A length that fits ONE clip of the render model renders as a single continuous pass; anything longer is STITCHED from acts filled to that model’s clip maximum with the remainder last (on a 15s-clip model, 40 → 15+15+10 and 17 → 13+4) — never time-compressed. The maximum is the model’s own: 15s on most, 30s on the longest-clip model, so a 30s ad can be one unbroken take rather than two acts. Omit when the user named no length; do NOT pass a guess, an omitted value keeps the recipe-aware default.",
      +  "type": "number"
      +}
    • addedInput schema / properties / hook
      Added value: +{
      +  "description": "force the VISUAL scroll-stop mechanic the opening beat is built on — a hook id from list_hooks (e.g. \"direct_callout\", \"mid_problem\", \"macro_asmr\"). Omit to let the planner pick. A hook that cannot be delivered in this brief is DROPPED with the reason rather than rendered wrongly — an on-screen-text hook on an authentic/UGC ad is the one that bites, because that register carries zero on-screen text.",
      +  "type": "string"
      +}
    • addedInput schema / properties / setting
      Added value: +{
      +  "description": "force the WHERE — a setting id from list_hooks (e.g. \"kitchen\", \"gym\", or a surreal one like \"volcano_rim\" / \"airplane_wing\", which are played 100% straight and never acknowledged). Omit for a neutral setting.",
      +  "type": "string"
      +}
    • changedOutput schema / (root)
      Previous value: -{
      -  "$schema": "http://json-schema.org/draft-07/schema#",
      -  "additionalProperties": false,
      -  "properties": {
      -    "brand": {
      -      "description": "the brand grounding embedded in the creative (name, logo, palette, productImages)"
      -    },
      -    "concept": {
      -      "description": "the one-line creative concept",
      -      "type": "string"
      -    },
      -    "copy": {
      -      "description": "copy variants ({headline, primary, cta})",
      -      "items": {},
      -      "type": "array"
      -    },
      -    "format": {
      -      "description": "the resolved creative format — 'image' or 'video'",
      -      "type": "string"
      -    },
      -    "image_concept": {
      -      "description": "the render-ready image concept (prompt etc.) when format is image"
      -    },
      -    "imodel": {
      -      "description": "the image model id to render with",
      -      "type": "string"
      -    },
      -    "recipe": {
      -      "description": "the resolved recipe id",
      -      "type": "string"
      -    },
      -    "recipe_label": {
      -      "description": "the resolved recipe display name",
      -      "type": "string"
      -    },
      -    "render_plan": {
      -      "description": "the routing plan (structure/duration) render_ad honors"
      -    },
      -    "video_storyboard": {
      -      "description": "the timed storyboard (scenes, cta, music) when format is video"
      -    },
      -    "vmodel": {
      -      "description": "the video model id to render with",
      -      "type": "string"
      -    }
      -  },
      -  "type": "object"
      -}New value: +null
  5. Addedv0.1.15
  6. Removedv0.1.0
  7. First observedv0.1.0

TDQS

A4.6/5.0
Behavior5/5

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

Discloses rich behavioral traits beyond the annotations: spends credits, renders nothing, clamps durationSeconds to 4–180, stitches long segments instead of time-compressing, drops undeliverable hooks with reasons, and notes the UGC on-screen-text pitfall. The annotations provide no safety hints (readOnlyHint false, destructiveHint false), so this added context carries the behavioral burden and does it thoroughly.

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

Conciseness4/5

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

The description is long but dense and every paragraph earns its place given the tool's complexity and 10 parameters. The core purpose is front-loaded and key behavioral warnings are highlighted with caps. Some redundancy exists between the opening 'length is sovereign' paragraph and the durationSeconds parameter description, but overall it is well-organized.

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?

Given no output schema, the description explains the output concept (copy variants, image prompt or storyboard, resolved recipe, model ids) and references supporting tools like get_brand, draft_brand, list_hooks, and hermoso_capabilities for parameter values. It covers prerequisites and billing, though a slightly more explicit return-shape description would make it fully complete.

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

Parameters5/5

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

Schema coverage is 100%, so the baseline is 3, but the description adds critical semantics beyond the schema: hook dropping behavior, brand profile vs saved brand, draft usage only after a video refusal, verbatim duration copying with clamping/stitching rules, and reference video watching behavior. This substantially exceeds what the JSON Schema alone provides.

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 description states a specific verb and resource: turning a brand + product/brief into a finished ad concept with copy variants and either an image prompt or video storyboard, plus recipe and model ids. It also differentiates from downstream render tools by explicitly saying 'Renders nothing' and from clone_video for remaking an existing video at its own length.

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?

Provides clear context: output should be chained into generate_image/generate_video, clone_video is named as the direct alternative for remaking one video at its own length, and render_ad is described as handling one clip or stitched acts. It does not enumerate all possible alternative planning tools, but the guidance is concrete and actionable.

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

Deploy Server

Other Tools