Segmently Product CLI Asset Slot Workflow
Strategy screen asset slot commands operate on existing screen-owned slots. The public Product CLI can add prompt and image experiments to a slot, regenerate a slot prompt, generate a new image, and select an experiment. It does not create brand-new asset slots.
1. Inspect available slots
List slots for a strategy block, then inspect a specific slot. Always pass
--screen-id for slot-level operations because the same slotKey can appear on
multiple screens.
segmently strategies blocks assets slots list <strategyId> <blockId> [projectId]
segmently strategies blocks assets slots get <strategyId> <blockId> <slotKey> --screen-id <screenId> --output slot.json [projectId]2. Generate the first prompt experiment
Use regenerate-prompt even when the slot has no usable prompt yet. Request
three concepts for normal review workflows.
segmently strategies blocks assets slots regenerate-prompt \
<strategyId> <blockId> <slotKey> \
--screen-id <screenId> \
--instruction-file asset-instruction.md \
--concept-count 3 \
--wait \
[projectId]When measured screen context is available, include it so the prompt designer can preserve the intended aspect ratio and slot fit:
segmently strategies blocks assets slots regenerate-prompt \
<strategyId> <blockId> <slotKey> \
--screen-id <screenId> \
--instruction-file asset-instruction.md \
--screen-context-file asset-screen-context.json \
--concept-count 3 \
--wait \
[projectId]3. Regenerate or repair from an existing prompt experiment
Use --source-experiment-id when revising a selected prompt experiment or
repairing a weak prompt. Use --select only when the new draft should become
the selected prompt experiment.
segmently strategies blocks assets slots regenerate-prompt \
<strategyId> <blockId> <slotKey> \
--screen-id <screenId> \
--source-experiment-id <experimentId> \
--instruction "Make the visual less literal" \
--concept-count 3 \
--select \
--wait \
[projectId]For targeted repair, add a compact issue report:
segmently strategies blocks assets slots regenerate-prompt \
<strategyId> <blockId> <slotKey> \
--screen-id <screenId> \
--source-experiment-id <experimentId> \
--source-package-id <packageId> \
--issue-report asset-issue-report.json \
--instruction-file asset-instruction.md \
--concept-count 3 \
--wait \
[projectId]4. Generate a new image experiment
Generate from a stored prompt experiment by passing --source-experiment-id.
Add public reference image URLs only when the operator explicitly selected them.
segmently strategies blocks assets slots generate \
<strategyId> <blockId> <slotKey> \
--screen-id <screenId> \
--source-experiment-id <promptExperimentId> \
--model gemini-3.1-flash-image-preview \
--aspect-ratio 9:16 \
--reference-images-file asset-reference-images.json \
--select \
--wait \
[projectId]Use explicit prompt overrides only for manual experiments:
segmently strategies blocks assets slots generate \
<strategyId> <blockId> <slotKey> \
--screen-id <screenId> \
--prompt-file image-prompt.md \
--provider google \
--model gemini-3.1-flash-image-preview \
--aspect-ratio 9:16 \
--select \
--wait \
[projectId]5. Select an existing experiment
Select a generated experiment when it has already been reviewed:
segmently strategies blocks assets slots select <strategyId> <blockId> <slotKey> <experimentId> --screen-id <screenId> [projectId]Verification
--waitreturns the terminal backend task status. Without--wait, poll withsegmently tasks get <taskId> --project <projectId>.- Run
assets slots getafter every write to confirmselectedExperimentId, prompt text, generated URL/status, and experiment count. - If generation fails because of project access, read the structured
requiredCapabilityvalue and stop until access is enabled.
Access
- Read commands use
projects:read. - Prompt/image mutations use
generate:write. - Waiting or polling tasks requires
tasks:read. - Project access is web onboarding access.