ai image repair prompt doctor
Diagnose failed AI images and generate surgically precise repair prompts for Midjourney, DALL-E, and Stable Diffusion.
$5
Works with the AI tools you already use
Claude Code
Cursor
Codex CLI
GitHub Copilot
Gemini CLI+20 moreClaude CodeCursorCodex CLIGitHub CopilotGemini CLI
OpenClaw
Manus
Grok+17 agentsai image repair prompt doctor
Example session with this skill installed
My prompt was 'cyberpunk street, neon signs, rainy, photorealistic' but I got a sunny day in a clean city with no neon. I want a dark, rainy, neon-drenched alley. Fix this for Midjourney.
- Read your context and instructions
- Compiled the ai image repair
- Generated the document
I have analyzed the failure of your cyberpunk prompt. The model defaulted to a standard city aesthetic because the lighting and time-of-day constraints were too vague to override the 'street' default.
Diagnosis Report
Summary
The original prompt failed because 'cyberpunk' and 'neon' were treated as style tags rather than environmental constraints. The model prioritized 'street' and 'photorealistic', defaulting to daylight. We need to explicitly kill the daylight and force a dark environment.
Intended vs Actual Comparison
| # | Prompt Claim | Actual Result | Status |
|---|---|---|---|
| 1 | Cyberpunk street | Clean modern city | Broken |
| 2 | Neon signs | None visible | Absent |
| 3 | Rainy | Dry pavement | Absent |
| 4 | Photorealistic | Sharp, clear | Kept |
Failure Classification
Claim 2 & 3 (Neon/Rain): Prompt Defect. The prompt lacked lighting hierarchy. Without 'night' or 'darkness', the model used a high-sun default.
Preserve / Correct / Remove / Constrain
- Preserve: Photorealistic texture.
- Correct: Lighting (change from sunny to midnight).
- Remove: Daylight, clean surfaces.
- Constrain: Rain intensity and neon color frequency.
Repair Prompt
cyberpunk alleyway at midnight, heavy rain, puddles reflecting neon light, glowing pink and cyan signage, cinematic lighting, dark atmosphere, ultra-detailed, photorealistic --ar 16:9 --no sunlight, daytime
Next steps
- Run this in Midjourney v6 for better adherence to the
--noparameters. - If the rain is too light, add weights like
heavy rain::2. - Use a specific street-level camera angle if the composition is still too wide.
Validation Check
- Clarity: Unambiguous night scene.
- Consistency: No conflicting light sources.
- Controllability: Specific colors and time defined.
- Completeness: Fixes rain and neon absence.
- No regressions: Maintains photorealism.
- Model fit: Uses Midjourney parameter syntax.
ai-image-repair-prompt-doctor.pdf
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
What you get
About this skill
The problem
AI image generators often ignore specific instructions, hallucinate unwanted elements, or produce anatomical errors. Developers and designers waste hours "prompt guessing" without understanding why a specific model failed to interpret their input.
What it does
- Deconstructs original prompts into subject, style, composition, and technical claims to find the point of failure.
- Maps intended vs. actual results in a structured comparison table to isolate "absent" or "broken" instructions.
- Classifies failures as either Prompt Defects or Model Limitations to prevent impossible repair attempts.
- Generates a surgically precise repair prompt using affirmative, weighted language and negative constraints.
- Validates the new prompt against a 6-point checklist to ensure clarity, consistency, and model controllability.
Frameworks & tools
Midjourney, DALL-E 3, Stable Diffusion, Flux, Adobe Firefly, and Ideogram.
Why this beats prompting it yourself
Instead of just adding more keywords, this skill performs a root-cause analysis. It identifies which parts of your prompt are working so you don't break them while fixing errors, and it stops you from fighting architectural model limits that a prompt cannot solve.
Use cases
- Fixing composition errors where the background overpowers the main subject.
- Correcting "style bleed" where an aesthetic instruction is ignored by the model.
- Removing unwanted subjects or artifacts without using vague negative prompts.
- Refining lighting and color palettes when the model defaults to high-saturation presets.
Known limitations
Cannot fix inherent model architectural failures like finger counts or text rendering through prompting alone. Requires the original prompt, failed image, and intended outcome to function.
How to install
Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.
- 1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
- 2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
- 3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.
Skills folder by agent
Click the path to copy it. Create the folder if it does not exist yet.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
30-day refund
Not a fit? Get your money back
Trust & safety
Security scanned
Verified clean 10 days ago
- Passed all security checks, Safe to install