image-factory

Generates article cover images with 5 dimensions (type, palette, rendering, text, mood) combining 11 color palettes and 7 rendering styles. Supports cinematic (2.35:1), widescreen (16:9), and square (1:1) aspects. Use when user asks to "generate cover image", "create article cover", or "make cover".

AI image generation with OpenAI GPT Image 2.5, Azure OpenAI, Google, OpenRouter, DashScope, Z.AI GLM-Image, MiniMax, Jimeng, Seedream, Replicate and Agnes APIs. Supports text-to-image, reference images, aspect ratios, and batch generation from saved prompt files. Sequential by default; use batch parallel generation when the user already has multiple prompts or wants stable multi-image throughput. Use when user asks to generate, create, or draw images.

Generates infographic image card series with 12 visual styles, 8 layouts, and 3 color palettes. Breaks content into 1-10 cartoon-style image cards optimized for social media engagement. Use when user mentions "小红书图片", "小红书种草", "小绿书", "微信图文", "微信贴图", "image cards", "图片卡片", baoyu-xhs-images, or wants social media infographic series.

Route image work between direct Baoyu generation and governed Image Factory batches, preserve each workflow's real execution boundary, and normalize artifact evidence for delivery or video handoff. Use when selecting an image workflow, handling a request that mixes creative generation with approval or receipts, or preparing image outputs for downstream media work.

Use when a batch has been generated and its results need a verdict, when the user asks whether a batch came out right, or when the results should be turned into another round. Runs the deterministic gates, records any advisory score next to the human labels so the two can later be compared, and writes the next round's plan from the items that actually need rework. Use when the user says "check these", "score the batch", or "try again with better prompts". For producing the images in the first place use `image-factory-run`.

Use when an image batch was interrupted, when a ledger reports a state the user does not understand, or when the user asks what happened to a batch and whether it is safe to continue. Reads the job ledger, maps the current state to the single legal next step, and reports it without spending anything. Use when the state is `Running`, `Unknown`, or `Partial`, or when a previous attempt stopped unexpectedly. For a batch that has not been started use `image-factory-run`.

Use when a governed Factory round needs an independent dimensioned critique of its generated images, when convergence across rounds must be shown with named-dimension evidence, or when a single opaque advisory score cannot direct the next rewrite. Writes the machine-readable advisory document that `evaluate` and `optimize` consume; the user-facing verdict stays with `image-factory-judge`.

Use when a batch plan exists and its images have not been produced yet, or when the user asks to generate the images in a plan. Validates the plan, quotes the batch, obtains approval, generates one image per item through Codex, and collects a hash-verified receipt for every artifact. Use when the user says "run this batch" or "generate these". For evaluating results that already exist use `image-factory-judge`; for a run whose state is unclear use `image-factory-recover`.

Use when the user wants to produce a batch of images, run an image production round, or continue an image batch that already exists. Routes to `image-factory-run` for a batch that has not been generated yet, to `image-factory-judge` for a batch with results to evaluate or a next round to plan, and to `image-factory-recover` for a batch whose state is unclear or that was interrupted. This skill only picks the entry point and never performs the work itself. For a single ad-hoc image with no batch plan, no skill in this plugin applies — ask Codex for the image directly.