zengwenliang416/skills-hubs/image-api-workbench

Image API Workbench

Purpose

Run reproducible Images API generation or editing against OpenAI or a reviewed compatible gateway. The CLI manages named gateway profiles and validates requests, catalogs, streaming partials, outputs, and evidence. Default to gpt-image-2.5-sunburst; choose gpt-image-2.5-flare when generation speed and throughput matter more than maximum edit precision. Catalogs are route evidence, not endpoint proof.

Workflow

1. Read model support when changing models or

advanced parameters.

1. Use --show-config for gateway diagnosis. Configure a named profile only

when a route needs to be persisted; never put an API key in profile JSON.

1. Use --dry-run for new or uncertain parameter combinations, and

--list-remote-models only when catalog discovery is needed.

1. Generate with /v1/images/generations; add --input-image to switch to

/v1/images/edits. Use transparent PNG/WebP or streaming flags when the selected endpoint supports them.

1. Save assets outside this skill and inspect the image plus metadata sidecar.

Both GPT Image 2.5 models accept xhigh and max quality. Keep gpt-image-2 only when an existing route explicitly requires it.

Safety Defaults

• Deprecated or retired models require an explicit allow flag; keep the current

model status in model support.

• Retired DALL-E models are blocked unless --allow-retired-model is explicit.
• Known GPT Image parameters are model-validated. Provider extensions require

--allow-provider-extensions.

• Credentials come from environment variables or token files. Never persist

keys in prompts, metadata, skill files, reports, or logs.

• Profile files persist only endpoint, model, timeout, and token-file path.

They are written with 0600 permissions.

• HTTP 524 means an intermediary gateway stopped waiting for its upstream;

increasing only the CLI timeout does not extend that gateway's own limit.

• Metadata stores a prompt hash by default. Prompt preview is opt-in.

Editing And Localization

For matched bilingual visuals, create one approved master and edit that exact image for each language. Preserve composition and non-text details; replace visible copy only. Build a side-by-side review with scripts/bilingual_contact_sheet.py.

See operating guide for commands, error classification, streaming behavior, and verification. Routing boundaries are covered by evals/trigger_cases.json.

Output Contract

A successful paid call requires:

• stdout JSON has ok: true;
• final image exists and is non-empty;
• metadata sidecar exists unless explicitly disabled;
• detected format and dimensions are plausible;
• actual_size_matches_request is checked rather than inferred;
• edit metadata identifies each input by path, bytes, SHA-256, and dimensions.

Catalogs and dry runs do not prove endpoint support. Paid probes require user approval.