Govern Atlas Cloud media generation through the registered project tools.
SkillScore70Maintained
Automatic
◆Marketing & Content pack
Last verified 8/17/2026
What it changes
Passed 14 of 20 test cases with the skill, and 5 of 20 without it.
Without skill26%
With skill70%
Checks it passed that the model alone missed
Declares image editing as unsupported in Atlas Cloud.
Loads ai-video-gen for general video generation guidance.
Uses Seedance operations through the atlascloud_video tool.
and 7 more in the full report
10 fixed6 passed either way7 still failing0 broke
Pass-rate lift
+43 pp
Tokens
-22%
Agent turns
0%
Response time
-50%
Model gemini-3.6-flashMeasured 8/17/2026v116 of 23 cases passed with skill · 20 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 10
Prompt
We have an existing product photo in 'product.png' and want to perform object removal and background replacement in Atlas Cloud. Please run the image editing workflow to alter the picture.
What was checked, without and with the skill
Checked
Without skill
With skill
Declares image editing as unsupported in Atlas Cloud.
Missed
Passed
case-04 · gemini-3.6-flash · 8/17/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md43 linessingle file
Atlas Cloud Media Generation
Use the registered atlascloud_image and atlascloud_video tools. Do not call
the provider API directly or invent model identifiers that are absent from the
tool metadata.
Workflow
Confirm that Atlas Cloud is the selected provider and that
ATLASCLOUD_API_KEY is available without printing or logging it.
Read the selected tool's live input schema and model_options.
Choose the operation and provide a project-scoped output_path.
Review the tool's cost estimate before generation.
Execute through the project pipeline and preserve its checkpoints.
Inspect the returned artifact and metadata before reporting success.
Image generation
Use atlascloud_image for the supported Seedream model.
Provide a prompt, one of the declared sizes, and an output format matching
the output filename.
Do not attach FLUX-, Wanx-, or other provider-specific guidance.
Treat image editing, unsupported models, and seed control as unsupported
unless the live tool schema changes.
Video generation
Use atlascloud_video for the declared Seedance operations.
Select text_to_video, image_to_video, or reference_to_video before
choosing inputs.
Supply the required first-frame or reference inputs for conditioned
operations.
Use only the URL, Base64, or asset-reference forms accepted by the live
schema; do not assume that a local image output path is uploadable.
Also load ai-video-gen for general video-generation guidance.
Guardrails
Let the tools manage asynchronous submission, polling, downloads, and output
validation.