Plan and validate a creative request BEFORE generating anything.

Use when: You want to preview costs, workflow, and fidelity before generating. This is the front door for planned creative workflows.

Do not use when: You want a quick one-shot generation (use canvas_create_image with goal directly), render HTML layouts (use canvas_render_layout), edit existing images, or manage brand assets.

Two actions:

• action="plan": provide a goal → get a CreativePlan with plan_id, PlanResultStatus

• action="approve": provide plan_id → approve the plan for execution

Example: canvas_plan_image(action="plan", goal="Instagram ad for handbag", subject_type="product", reference_image="/path/to/bag.jpg")

Generate marketing images — plan-based or direct mode.

Use when: You want to generate marketing images. Either pass a plan_id from canvas_plan_image (recommended for complex jobs), or pass goal + details directly for quick one-shot generation.

Do not use when: You want to render HTML layouts (use canvas_render_layout), edit existing images (use canvas_edit_image), or analyze images (use canvas_inspect_asset).

IMPORTANT: After this tool returns, ALWAYS open the generated image(s) in Preview (macOS) or the system image viewer so the user can review the output. Use: open -a Preview

Example (plan-based): canvas_create_image(goal="Instagram ad", plan_id="plan_abc") Example (direct): canvas_create_image(goal="Hero banner for website", target_channels=["og_image"], subject_type="scene")

Render HTML/CSS templates or raw HTML to pixel-perfect marketing images.

Use when: The user explicitly asks for a designed layout, template render, banner with specific typography, or HTML/CSS composition. This is a deterministic renderer, NOT AI generation.

Do not use when: The user wants AI-generated photography or illustrations. For those, use canvas_create_image (direct or plan-based).

Example: canvas_render_layout(template="hero_banner", variables={"headline": "Launch Day"}, brand_kit_id="acme")

IMPORTANT: After this tool returns, ALWAYS open the rendered image(s) in Preview (macOS) or the system image viewer so the user can review the output. Use: open -a Preview

Modify, enhance, or transform an existing image.

Use when: The user has an image and wants to change it — edit content, remove background, upscale resolution, or apply color grading.

Do not use when: The user wants to create something new (use canvas_create_image) or render an HTML layout (use canvas_render_layout).

Actions:

• edit: Text-directed image editing (requires prompt)

• remove_background: Isolate subject with transparent PNG

• upscale: 2-4x resolution enhancement via Real-ESRGAN

• apply_grading: Apply color grading preset to image

Example: canvas_edit_image(image="/path/to/photo.png", action="edit", prompt="make the sky sunset orange")

IMPORTANT: After this tool returns, ALWAYS open the output image in Preview (macOS) or the system image viewer so the user can review the result. Use: open -a Preview

Create, list, search, and inspect brand kits and generated assets.

Use when: The user wants to set up brand identity, manage logos, colors, fonts, inspect brand kits, or search for previously generated assets.

Do not use when: The user wants to generate or edit images.

Brand kit actions: create, list, get, delete Generated asset actions: list, get, search, delete

Example: canvas_manage_library(target="brand_kit", action="create", name="acme", color_palette=["#1a1a2e"], visual_vibe="modern minimal")

Analyze an image using AI vision — describe contents, extract attributes, QA check.

Use when: The user wants to understand an image, verify product fidelity, extract structured attributes, or QA check a generated asset.

Do not use when: The user wants to generate or modify images.

Example: canvas_inspect_asset(image="/path/to/bag.png", question="Does this match the original product shape?")