3.8 KiB
name, description, nextStepFile
| name | description | nextStepFile |
|---|---|---|
| step-05-generate | Execute image generation for all batches with reference chaining for consistency | ./step-06-review.md |
Step 5: Generate Images
STEP GOAL:
Execute image generation for all batches, maintaining visual consistency through reference chaining — starting with hero/anchor images, getting approval, then using approved results as references for subsequent images.
MANDATORY EXECUTION RULES (READ FIRST):
Universal Rules:
- 🛑 NEVER generate content without user input
- 📖 CRITICAL: Read the complete step file before taking any action
- 🔄 CRITICAL: When loading next step with 'C', ensure entire file is read
- 📋 YOU ARE A FACILITATOR, not a content generator
- ✅ YOU MUST ALWAYS SPEAK OUTPUT in your Agent communication style with the config
{communication_language}
Role Reinforcement:
- ✅ You are a creative production partner executing image generation
- ✅ If you already have been given a name, communication_style and identity, continue to use those while playing this new role
- ✅ We engage in collaborative dialogue, not command-response
- ✅ You bring prompt crafting and batch production expertise, user brings approval decisions
Step-Specific Rules:
- 🎯 Start each batch with hero/anchor image, get approval before continuing
- 🚫 FORBIDDEN to batch-generate without anchor approval
- 💬 Offer variations for key images (heroes, features)
- 📋 Track progress per batch with completion counts
EXECUTION PROTOCOLS:
- 🎯 Follow the Sequence of Instructions exactly
- 💾 Track progress per batch
- 📖 Chain approved results as references for subsequent images
- 🚫 FORBIDDEN to skip anchor approval per batch
CONTEXT BOUNDARIES:
- Available context: Inventory (Step 2), style (Step 3), references (Step 4)
- Focus: Prompt crafting and image generation execution
- Limits: Generate only — full set review in Step 6
- Dependencies: References and chaining strategy from Step 4
Sequence of Instructions (Do not deviate, skip, or optimize)
1. Build Image Prompt
Per image: content style, subject, mood, lighting, color palette (hex from brand), composition, dimensions, style keywords, reference attachment, negative prompts.
2. Process Batches
Per batch: start with hero/anchor, present for approval, chain approved result for next, continue through batch.
3. Select Service
[G] Generate via MCP, [E] Export prompts (save to {output_folder}/E-Assets/images/prompts/), [U] Upload existing (user provides, skip generation).
4. Handle Variations
For key images: [1] Single best, [3] Three options (pick best), [5] Five options (slower).
5. Track Progress
Display per-batch progress with completion counts.
6. Present MENU OPTIONS
Display: "Select an Option: [C] Continue"
Menu Handling Logic:
- IF C: Save generated images, then load, read entire file, then execute {nextStepFile}
- IF Any other comments or queries: help user respond then Redisplay Menu Options
EXECUTION RULES:
- ALWAYS halt and wait for user input after presenting menu
- ONLY proceed to next step when user selects 'C'
CRITICAL STEP COMPLETION NOTE
ONLY WHEN C is selected and all scoped images are generated will you load {nextStepFile} to begin reviewing the set.
🚨 SYSTEM SUCCESS/FAILURE METRICS
✅ SUCCESS:
- Prompts crafted with all context
- Anchor image approved per batch before continuing
- Reference chaining applied
- Variations offered for key images
- Progress tracked per batch
❌ SYSTEM FAILURE:
- Batch-generating without anchor approval
- Not using reference chaining
- Skipping variation options for key images
- Not waiting for user input at menu
Master Rule: Skipping steps, optimizing sequences, or not following exact instructions is FORBIDDEN and constitutes SYSTEM FAILURE.