remove-background-workflowlisted
Install: claude install-skill devkindhq/ideogram-ai-toolkit
# Remove Background Workflow
This skill orchestrates `remove_background` as its core call, pulling in `upload_image`,
`get_recent_generations`, and `get_generation_status` as supporting tools depending on how
the source image gets resolved. It's a thin orchestration workflow around a single MCP
call, not a prompt-composition skill — there's no `composition-spec-format.md`/
`panel-anatomy.md`-style reference here.
This skill handles exactly one image per invocation. If the user asks for several images at
once, run the workflow below once per image and report each result separately —
`remove_background` itself only accepts one source per call.
## Before you start
Read `references/background-removal-patterns.md` before running step 2 of the workflow
below. It covers the single-source rule (why `image_response_id` and `image_upload_id` can
never both be set), the private/public default, and how to handle the running/pending
envelope some calls return instead of an immediate result.
## Workflow
### 1. Resolve the source image
Figure out what the user means by "this image" before calling anything:
- **A response from earlier this session** — if the user is pointing at something just
generated or edited, pull the `response_id` from `structured_content.response_ids` on the
prior `generate_image`/`edit_image`/etc. call.
- **A freshly attached local file** — run the `upload_image` flow first to mint an
`image_upload_id`: call `upload_image(filename=...)`, run the returned