Midjourney

Build me a local image-generation studio wrapping open models, in place of
Midjourney. Requirements:

- A local web app (Node + Express, plain HTML/JS) on localhost:4890: prompt box,
  a generate-4 button, and a gallery grid of results.
- Backend targets whichever engine I have: a local ComfyUI instance running Flux
  or SDXL if there is a GPU, otherwise a hosted API like Replicate or fal.ai
  (key in .env). Same UI either way.
- Save every image to ~/ImageGen/YYYY-MM/ with a sidecar JSON: prompt, seed,
  model, steps, so any result is reproducible later.
- Gallery search over past prompts via SQLite FTS5, favorites, and one-click
  re-run with the same seed or a new variation.
- Style presets in a styles.json of prompt prefixes and suffixes I can toggle,
  a poor man's style reference.
- No accounts, no telemetry, everything stays local except the hosted-API calls.
- Out of scope: training or fine-tuning models, and matching Midjourney's look.
  Build for reproducibility and control, not a taste layer.
- README: both engine setups, the rough cost per image on the hosted route, and
  a plain note that this is not Midjourney, the proprietary model and its years
  of aesthetic tuning cannot be rebuilt, this trades that for privacy and control.

# AGENTS.md — Build guide for Midjourney

## Project scope
Use Stable Diffusion/Flux-style local or hosted models with a prompt UI and gallery.

Catalogue verdict: no. You can build a wrapper around open models, but you cannot solo-recreate Midjourney's proprietary model quality, style tuning, infrastructure, and community distribution.
Use the implementation prompt below to define the deliverable. Complete each phase's acceptance checks before extending the scope.

## Working agreement
- Inspect the repository and its existing instructions before choosing paths, dependencies or commands. Keep one coherent stack and explain changes to the proposed architecture.
- Plan a vertical slice that accepts a real input and produces the useful output described below. Persist only the state the prompt calls for; respect memory-only and upstream-managed workflows. Use fixtures only when they are clearly labelled.
- After scaffolding, document the actual install, development, check and build commands in README and keep them synchronized with the package or project manifest. Do not report commands as successful unless they ran.
- Work in small steps. At handoff, list implemented flows, checks actually performed, remaining blockers, and any credentials or provider setup the owner must supply.
- Do not publish, spend money, contact customers, delete source data or run irreversible migrations without the project owner's authorization.

## Prerequisites
- GPU or hosted image API
- model weights/license
- storage/gallery
- prompt UI

## Stack and architecture
- A local web app (Node + Express, plain HTML/JS) on localhost:4890: prompt box, a generate-4 button, and a gallery grid of results.
- Model prompts, model settings, generation jobs, source assets, and exported images; keep stable source IDs and timestamps.
- Interface for this image generation workflow: a local web page with input, progress, review, and export views.

## Security and data integrity
- Keep the original media file untouched when decoding or rendering fails; show the failing job and allow retry.
- Validate imported records and file paths; keep user data within the storage location named in the prompt.
- State the practical limit: frontier/proprietary model quality.

## Agent implementation rules
- Project rule, data: Model prompts, model settings, generation jobs, source assets, and exported images; keep stable source IDs and timestamps.
- Project rule, behavior: Backend targets whichever engine I have: a local ComfyUI instance running Flux or SDXL if there is a GPU, otherwise a hosted API like Replicate or fal.ai (key in .env). Same UI either way.
- Project rule, recovery: Keep the original media file untouched when decoding or rendering fails; show the failing job and allow retry.

## Optional agent skills and references
No installable skill is listed for this guide; use the project rules and the stack documentation.

Read the linked SKILL.md and its dependencies before adding a skill. Select only the skills matching this project's runtime and task; their documentation does not supply API access, credentials or approval to perform external actions. Pin the reviewed revision where the tool supports it. Follow the chosen agent's documented project-level installation mechanism.

## Distribution ideas
These are optional planning notes. Obtain the owner's approval before publishing or contacting anyone.
- Demonstrate this image generation operation with sample or user-owned data: Backend targets whichever engine I have: a local ComfyUI instance running Flux or SDXL if there is a GPU, otherwise a hosted API like Replicate or fal.ai (key in .env). Same UI either way.
- Show a concrete result or export: Text-to-image generation service known for high-quality aesthetic output.
- State the practical limit: frontier/proprietary model quality.

## Engineering roadmap
1. Phase 1, architecture and data: A local web app (Node + Express, plain HTML/JS) on localhost:4890: prompt box, a generate-4 button, and a gallery grid of results. Model prompts, model settings, generation jobs, source assets, and exported images; keep stable source IDs and timestamps.
2. Phase 2, implement: Backend targets whichever engine I have: a local ComfyUI instance running Flux or SDXL if there is a GPU, otherwise a hosted API like Replicate or fal.ai (key in .env). Same UI either way.
3. Phase 3, implement: Use Stable Diffusion/Flux-style local or hosted models with a prompt UI and gallery.
4. Phase 4, review and output: Text-to-image generation service known for high-quality aesthetic output.
5. Phase 5, recovery and acceptance: Keep the original media file untouched when decoding or rendering fails; show the failing job and allow retry. Verify this invariant with a saved fixture: A failed or cancelled model job retains its prompt and source assets; an exported image must reopen at the recorded dimensions. State the practical limit: frontier/proprietary model quality.

## Paid-product capabilities outside this build
- frontier/proprietary model quality
- style consistency
- moderation
- compute scale
- community/distribution
- model updates

## Implementation prompt
Build me a local image-generation studio wrapping open models, in place of
Midjourney. Requirements:

- A local web app (Node + Express, plain HTML/JS) on localhost:4890: prompt box,
  a generate-4 button, and a gallery grid of results.
- Backend targets whichever engine I have: a local ComfyUI instance running Flux
  or SDXL if there is a GPU, otherwise a hosted API like Replicate or fal.ai
  (key in .env). Same UI either way.
- Save every image to ~/ImageGen/YYYY-MM/ with a sidecar JSON: prompt, seed,
  model, steps, so any result is reproducible later.
- Gallery search over past prompts via SQLite FTS5, favorites, and one-click
  re-run with the same seed or a new variation.
- Style presets in a styles.json of prompt prefixes and suffixes I can toggle,
  a poor man's style reference.
- No accounts, no telemetry, everything stays local except the hosted-API calls.
- Out of scope: training or fine-tuning models, and matching Midjourney's look.
  Build for reproducibility and control, not a taste layer.
- README: both engine setups, the rough cost per image on the hosted route, and
  a plain note that this is not Midjourney, the proprietary model and its years
  of aesthetic tuning cannot be rebuilt, this trades that for privacy and control.

## Completion evidence
Demonstrate the prompt's acceptance scenarios against the scoped workflow. Include setup from a clean checkout and failure recovery. Check persistence across restart and export/restore only for the state the prompt says to store; for memory-only tools, confirm that temporary content is discarded as specified. Record actual results and remaining limitations. A detailed plan alone does not establish a working replacement.