Runner AI
DocumentationWork with Runner

Images, Models, and Skills

Add reference images and choose the Runner capabilities that fit a request.

What this covers: You will give Runner the right context and working tools before sending a request.

Images show Runner what you mean. Models and skills change the capabilities available for the next request. Use them deliberately; a clear request is still the most important input.

Entry path: Open a store project → conversation composer.

Before you start

  • Use only images you are allowed to share with the current workspace.
  • Remove secrets, payment data, and unnecessary customer information.
  • Check plan and browser access when a model, Skill, or screen-capture action is missing.

Composer controls

ControlWhat happens when you select it
AddOn desktop, opens Upload image and, when supported, Take screenshot.
Upload imageOpens the image picker for JPG/JPEG, PNG, GIF, WebP, BMP, or TIFF files.
Take screenshotStarts screen capture when the browser and account support it. The browser may ask for permission.
Picture on mobileOpens the mobile image picker. It is the only upload choice in the mobile sheet.
Model selectorLists the Runner models available to the account. A locked model opens the relevant Plan & Billing path instead of selecting it.
Quick commands or SkillsOpens available shortcuts that can be added to the request.
More actionsReveals composer tools that do not fit in the current layout.

Runner ignores an image with the same name and size as one already in the draft and shows Duplicate image ignored. Unsupported files are ignored with Use a JPG, PNG, GIF, WebP, BMP, or TIFF image. In a mixed selection, supported images remain available. Runner currently accepts up to 10 images per message, up to 32 MB each.

Add an image

  1. Draft enough of the request to explain why the file matters.
  2. Select Add on desktop, or the upload button on mobile.
  3. Choose Upload image, Picture, or Take screenshot, depending on what is shown.
  4. Pick or capture the relevant image.
  5. Confirm that it appears with the draft.
  6. Remove it before sending if it is the wrong reference.
  7. Select Send.

If screen capture is unavailable or permission is denied, take a screenshot with your device and upload the saved image instead.

Choose a model

  1. Select the current model in the composer.
  2. Review the models Runner makes available.
  3. Choose the model that matches the task.
  4. Confirm the selected model is shown before you send.

If a model is locked, keep the current model or follow the plan information Runner opens. A locked choice is not applied to the draft.

Add a skill

  1. Open Quick commands, Skills, or the command palette.
  2. Search or browse for the capability you need.
  3. Select an already available skill to add or remove it from the request.
  4. Select a discoverable skill to add it when your plan permits.
  5. Return to the composer and confirm the intended skill is present.
  6. Send the request.

When the current plan does not include a skill, Runner can open the Agent or upgrade path instead. Memory choices appear only when memory is enabled for the account.

Troubleshooting

ProblemWhat to do
The image is not addedUse a supported image format, keep it under 32 MB, and keep the message at 10 images or fewer.
Duplicate image ignoredThe same name-and-size image is already in the draft. Keep the existing copy or remove it before selecting the image again.
Unsupported-file warningChoose a JPG/JPEG, PNG, GIF, WebP, BMP, or TIFF image. Videos, spreadsheets, and other files are not added.
Take screenshot is missingUse a saved image; live capture is conditional.
Screen capture stopsAllow the browser permission when appropriate, or attach a normal screenshot.
A skill will not toggleFinish any conflicting composer action and confirm that the plan includes the skill.
A locked model opens billingClose the plan screen to keep the current model, or review access with the workspace owner.