Xpersona
A

AI Image Generation

CLAWHUB · Unverified

Create AI images with GPT Image, Gemini Nano Banana, FLUX, Imagen, and top providers using prompt engineering, style control, and smart editing. Skill: AI Image Generation Owner: ivangdavila Summary: Create AI images with GPT Image, Gemini Nano Banana, FLUX, Imagen, and top providers using prompt engineering, style control, and smart editing. Tags: latest:1.0.3 Version history: v1.0.3 | 2026-03-02T11:45:32.596Z | user Updated for 2026 with benchmark-backed model selection and clearer guidance for modern image generation stacks. v1.0.2 | 2026-02-18T20:49:57.13

62
Rank
84
Safety
13k
Downloads

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 13.4K downloads reported by the source. Last updated 11/16/2026.

Avoid when

  • Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Install and run

Setup complexity: low.

clawhub skill install s178jdk12x4qj3gs2se3etxf3h83h7ft:image-generation
  1. Setup complexity is LOW. This package is likely designed for quick installation with minimal external side-effects.
  2. Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-ivangdavila-image-generation-neration-2/snapshot"

Run-check

$0.02 USD

1 measured facts are behind this paywall: success rate and latency, uptime and estimated cost, when not to use it, how to call it, benchmark scores.

Agents pay $0.02 in USDC. A card payment is $0.50, the smallest a card allows.

Documentation

CLAWHUB

97,807 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: AI Image Generation
slug: image-generation
version: 1.0.3
homepage: https://clawic.com/skills/image-generation
description: Create AI images with GPT Image, Gemini Nano Banana, FLUX, Imagen, and top providers using prompt engineering, style control, and smart editing.
changelog: Updated for 2026 with benchmark-backed model selection and clearer guidance for modern image generation stacks.
metadata: {"clawdbot":{"emoji":"🎨","requires":{"bins":[],"env.optional":["OPENAI_API_KEY","GEMINI_API_KEY","BFL_API_KEY","GOOGLE_CLOUD_PROJECT","REPLICATE_API_TOKEN","LEONARDO_API_KEY","IDEOGRAM_API_KEY"],"config":["~/image-generation/"]},"os":["linux","darwin","win32"]}}
---

## Setup

On first use, read `setup.md`.

## When to Use

User needs AI-generated visuals, edits, or consistent image sets.
Use this skill to pick the right model, write stronger prompts, and avoid outdated model choices.

## Architecture

User preferences persist in `~/image-generation/`. See `memory-template.md` for setup.

```
~/image-generation/
├── memory.md      # Preferred providers, project context, winning recipes
└── history.md     # Optional generation log
```

## Quick Reference

| Topic | File |
|-------|------|
| Initial setup | `setup.md` |
| Memory template | `memory-template.md` |
| Migration guide | `migration.md` |
| Benchmark snapshots | `benchmarks-2026.md` |
| Prompt techniques | `prompting.md` |
| API handling | `api-patterns.md` |
| GPT Image (OpenAI) | `gpt-image.md` |
| Gemini and Imagen (Google) | `gemini.md` |
| FLUX (Black Forest Labs) | `flux.md` |
| Midjourney | `midjourney.md` |
| Leonardo | `leonardo.md` |
| Ideogram | `ideogram.md` |
| Replicate | `replicate.md` |
| Stable Diffusion | `stable-diffusion.md` |

## Core Rules

### 1. Resolve aliases to official model IDs first

Community names shift quickly. Before calling an API, map the nickname to the provider model ID.

| Community label | Official model ID to try first | Notes |
|-----------------|--------------------------------|-------|
| Nano Banana | `gemini-2.5-flash-image-preview` | Common nickname, not an official Google model ID |
| Nano Banana 2 / Pro | Verify provider docs | Usually a provider preset over Gemini image models |
| GPT Image 1.5 | `gpt-image-1.5` | Current OpenAI high-tier image model |
| GPT Image mini / iMini | `gpt-image-1-mini` | Budget/faster OpenAI variant |
| FLUX 2 Pro / Max | `flux-pro` / `flux-ultra` | Many platforms rename these SKUs |

### 2. Pick models by task, not by hype

| Task | First choice | Backup |
|------|--------------|--------|
| Exact text in image | `gpt-image-1.5` | Ideogram |
| Multi-turn edits | `gemini-2.5-flash-image-preview` | `flux-kontext-pro` |
| Photoreal hero shots | `imagen-4.0-ultra-generate-001` | `flux-ultra` |
| Fast low-cost drafts | `gpt-image-1-mini` | `imagen-4.0-fast-generate-001` |
| Character/product consistency | `flux-kontext-max` | `gpt-image-1.5` with references |
| Local no-API workflows | `flux-schnell` | SDXL |

#

_meta.json

{
  "ownerId": "kn73vp5rarc3b14rc7wjcw8f8580t5d1",
  "slug": "image-generation",
  "version": "1.0.3",
  "publishedAt": 1772451932596
}

api-patterns.md

# API Patterns (2026)

Use provider-agnostic patterns to keep image workflows robust under model churn.

## 1) Alias Resolution Before API Call

Always normalize user/provider labels to official model IDs.

```python
MODEL_ALIASES = {
    "nano banana": "gemini-2.5-flash-image-preview",
    "nano banana pro": "gemini-2.5-flash-image-preview",
    "gpt image mini": "gpt-image-1-mini",
    "flux 2 pro": "flux-pro",
    "flux 2 max": "flux-ultra",
}
```

## 2) Tiered Fallbacks

```python
FALLBACKS = {
    "gpt-image-1.5": ["gpt-image-1", "gpt-image-1-mini"],
    "gemini-2.5-flash-image-preview": ["imagen-4.0-generate-001"],
    "flux-ultra": ["flux-pro", "flux-1.1-pro-ultra"],
}
```

If primary fails with `model_not_found` or permission errors, retry with the next tier.

## 3) Async Polling for Job APIs

```python
import time
import requests


def wait_for_job(url, headers, timeout_s=180, sleep_s=2):
    started = time.time()
    while time.time() - started < timeout_s:
        r = requests.get(url, headers=headers, timeout=30)
        data = r.json()
        status = data.get("status", "")
        if status in {"succeeded", "completed"}:
            return data
        if status in {"failed", "error", "cancelled"}:
            raise RuntimeError(data)
        time.sleep(sleep_s)
    raise TimeoutError("image job timed out")
```

## 4) Deterministic Request Logs

Store metadata for each generation:
- Provider
- Official model ID
- Prompt hash
- Seed (if supported)
- Cost tier
- Result URL/path

This makes quality and cost comparisons reproducible.

## 5) Cost Control Pattern

- First pass: draft tier (`mini`, `fast`, or local)
- Second pass: one focused edit on the winner
- Final pass: high-tier render only once

## 6) Webhook Pattern (When You Own the Receiver)

Some providers support webhooks for async completion.
Use webhooks only when you control the callback endpoint and audit logs.
Default to polling for simpler and safer setups.

## 7) Batch Concurrency Pattern

```python
import asyncio


async def generate_batch(prompts, max_concurrent=5):
    semaphore = asyncio.Semaphore(max_concurrent)

    async def _one(prompt):
        async with semaphore:
            return await async_generate(prompt)

    return await asyncio.gather(*[_one(p) for p in prompts])
```

## 8) Temporary URL Handling

Many providers return expiring URLs.
If output must persist, download immediately and store with request metadata.

## 9) Error Handling Baseline

```python
try:
    result = generate_image(prompt, model=model_id)
except RateLimitError:
    sleep_and_retry()
except PermissionError:
    model_id = fallback_model(model_id)
    result = generate_image(prompt, model=model_id)
except TimeoutError:
    simplify_prompt_and_retry()
```

## Provider Snapshot

| Provider | Typical Pattern | Notes |
|----------|------------------|-------|
| OpenAI GPT Image | Sync generate/edit | Model access varies by account tier |
| Gemini API | Conversational generate/edit | Previ

benchmarks-2026.md

# Benchmark Snapshot (2025-2026)

Use this file as a dated reference, not a permanent truth.

## Scope

- Source: LMArena Text-to-Image leaderboard (updated 2026-02-25)
- Source: LMArena Image Editing leaderboard (updated 2026-02-26)
- This reflects human preference votes, not absolute quality guarantees for every task

## Text-to-Image Snapshot (LMArena, 2026-02-25)

Top models shown on the board:
- #1 `GPT-image-1.5 High` (score 1162)
- #2 `Recraft V3` (1161)
- #3 `Imagen 4` (1157)
- #4 `GPT-image-1.5 Medium` (1153)
- #5 `Gemini 2.5 Flash Image` (1152)

Notable FLUX placements on the same board:
- `Flux Pro 1.1` (#9)
- `Flux 1.1 Pro` (#12)
- `Flux Kontext Max` (#14)
- `Flux Kontext Pro` (#16)

## Image Editing Snapshot (LMArena, 2026-02-26)

Top models shown on the board:
- #1 `Gemini 2.5 Flash Image` (1166)
- #2 `GPT-image-1.5 High` (1143)
- #3 `SeedEdit 3.0` (1142)
- #4 `Flux Kontext Max` (1137)
- #5 `Flux Kontext Pro` (1135)

## Practical Routing Based on These Results

- Text-heavy generation: start with `gpt-image-1.5`
- Iterative edits: start with `gemini-2.5-flash-image-preview`
- Edit quality with identity retention: test `flux-kontext-max` and `flux-kontext-pro`
- Fast ideation: use lower tiers (`gpt-image-1-mini`, `imagen-4.0-fast-generate-001`) before premium rerender

## Important Caveat

Leaderboard rank can shift quickly as providers tune models.
When output quality is critical, re-check rankings before final delivery.

flux.md

# FLUX (Black Forest Labs)

## Official BFL Model IDs

Use the model catalog from `docs.bfl.ai` and `api.bfl.ai`.
Current first-party IDs include:

| Model ID | Best For |
|----------|----------|
| `flux-pro` | Strong general text-to-image quality |
| `flux-ultra` | Higher-tier fidelity and detail |
| `flux-kontext-pro` | Image editing with context consistency |
| `flux-kontext-max` | Highest quality Kontext editing |
| `flux-kontext-dev` | Dev/testing for Kontext workflows |
| `flux-1.1-pro` | Stable prior-gen production path |
| `flux-1.1-pro-ultra` | Higher-fidelity 1.1 tier |
| `flux-dev` | Open/dev workflows |
| `flux-schnell` | Fast local/open baseline |

## About "FLUX 2"

Many third-party providers market FLUX tiers as "FLUX 2 Pro/Max".
Treat this as an alias and resolve to BFL IDs before API calls.

Typical mapping:
- "FLUX 2 Pro" -> `flux-pro`
- "FLUX 2 Max" -> `flux-ultra` or `flux-kontext-max` depending whether it is generation or editing

## API Examples

### Text-to-Image

```bash
curl -X POST "https://api.bfl.ai/v1/flux-pro" \
  -H "Authorization: Bearer $BFL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Product hero photo of matte black headphones on a marble table",
    "width": 1024,
    "height": 1024
  }'
```

### Kontext Edit

```bash
curl -X POST "https://api.bfl.ai/v1/flux-kontext-pro" \
  -H "Authorization: Bearer $BFL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Keep the product unchanged, replace background with warm studio lighting"
  }'
```

## Capability Notes

- FLUX families are strong for photoreal output and material detail
- Kontext tiers are useful when edits must preserve identity and structure
- FLUX is often used for consistent campaign assets across many scenes

## Local/Open Route

For local usage without paid APIs:
- `flux-schnell` for fast iteration
- `flux-dev` for deeper experimentation

Keep this route for budget workflows and private/offline development.

### Local Quick Start (Diffusers)

```python
import torch
from diffusers import FluxPipeline

pipe = FluxPipeline.from_pretrained(
    "black-forest-labs/FLUX.1-schnell",
    torch_dtype=torch.bfloat16,
)
pipe.enable_model_cpu_offload()

image = pipe(
    prompt="Editorial product photo of a watch on dark slate",
    width=1024,
    height=1024,
    num_inference_steps=4,
    guidance_scale=0,
).images[0]
```

## Model Selection Rules

- High quality generation: `flux-ultra`
- Balanced production generation: `flux-pro`
- Editing with source preservation: `flux-kontext-pro`
- Highest edit quality: `flux-kontext-max`
- Free/local drafts: `flux-schnell`

## Operational Tips

- For consistency jobs, lock prompt structure and aspect ratio across batches
- Use draft tiers for ideation, premium tiers for final delivery
- Keep alias-to-ID mapping explicit when switching providers

## Common Mistakes

- Calling `flux-2-pro` directly on BFL API (not an official endpoint)
- Using generation model IDs for 

Sponsored

Ads related to AI Image Generation and adjacent AI workflows.