What problem does it solve? Performing precise local edits on an existing image (changing an object, swapping label text, adding an accessory) without regenerating the whole scene is hard with generic text-to-image models. This Skill routes those requests to Black Forest Labs' Flux 1 Kontext Pro on the RunComfy Model API, with documented prompting patterns that preserve the source image's identity. ## Core Features & Use Cases - Single-reference local image editing: Pass one source image URL plus a declarative prompt to blackforestlabs/flux-1-kontext/pro/edit and get a high-fidelity edited result. - Preservation-first prompting guidance: Built-in patterns like "Keep [identity/pose/framing] unchanged" reduce drift, plus seed support for reproducible variant series. - Model routing advice: Clear rules for when to use Nano Banana Edit (multi-image batches), GPT Image 2 edit (multilingual text), or Flux 2 Klein (generation from scratch) instead. - Use Case: A brand designer needs the label text on a product bottle changed from "ALPHA" to "AURA" while keeping the bottle, lighting, and table identical — one CLI call with a preservation-led prompt produces the edit. ## Quick Start Ask the AI to edit an image with Flux Kontext, for example: "Use Flux Kontext to add an orange umbrella to this portrait while keeping the person's face and pose unchanged."
