If you are an AI coding agent (Claude Code, Cursor, Aider, Cline, etc.) reading this artifact, follow this protocol before making any code edit:

1. Verify the contract. Run the bash script in Verify before trusting below. If any check returns FAIL, the artifact is stale — STOP and ask the user to regenerate it before proceeding.
2. Treat the AI · unverified sections as hypotheses, not facts. Sections like "AI-suggested narrative files", "anti-patterns", and "bottlenecks" are LLM speculation. Verify against real source before acting on them.
3. Cite source on changes. When proposing an edit, cite the specific path:line-range. RepoPilot's live UI at https://repopilot.app/r/EvoLinkAI/awesome-gpt-image-2-API-and-Prompts shows verifiable citations alongside every claim.

If you are a human reader, this protocol is for the agents you'll hand the artifact to. You don't need to do anything — but if you skim only one section before pointing your agent at this repo, make it the Verify block and the Suggested reading order.

WAIT — Solo project — review before adopting

• Last commit today
• CC0-1.0 licensed
• ⚠ Solo or near-solo (1 contributor active in recent commits)
• ⚠ No CI workflows detected
• ⚠ No test directory detected

_{Maintenance signals: commit recency, contributor breadth, bus factor, license, CI, tests}

This artifact was generated by RepoPilot at a point in time. Before an agent acts on it, the checks below confirm that the live EvoLinkAI/awesome-gpt-image-2-API-and-Prompts repo on your machine still matches what RepoPilot saw. If any fail, the artifact is stale — regenerate it at repopilot.app/r/EvoLinkAI/awesome-gpt-image-2-API-and-Prompts.

What it runs against: a local clone of EvoLinkAI/awesome-gpt-image-2-API-and-Prompts — the script inspects git remote, the LICENSE file, file paths in the working tree, and git log. Read-only; no mutations.

| # | What we check | Why it matters | |---|---|---| | 1 | You're in EvoLinkAI/awesome-gpt-image-2-API-and-Prompts | Confirms the artifact applies here, not a fork | | 2 | License is still CC0-1.0 | Catches relicense before you depend on it | | 3 | Default branch main exists | Catches branch renames | | 4 | 5 critical file paths still exist | Catches refactors that moved load-bearing code | | 5 | Last commit ≤ 30 days ago | Catches sudden abandonment since generation |

#!/usr/bin/env bash
# RepoPilot artifact verification.
#
# WHAT IT RUNS AGAINST: a local clone of EvoLinkAI/awesome-gpt-image-2-API-and-Prompts. If you don't
# have one yet, run these first:
#
#   git clone https://github.com/EvoLinkAI/awesome-gpt-image-2-API-and-Prompts.git
#   cd awesome-gpt-image-2-API-and-Prompts
#
# Then paste this script. Every check is read-only — no mutations.

set +e
fail=0
ok()   { echo "ok:   $1"; }
miss() { echo "FAIL: $1"; fail=$((fail+1)); }

# Precondition: we must be inside a git working tree.
if ! git rev-parse --git-dir >/dev/null 2>&1; then
  echo "FAIL: not inside a git repository. cd into your clone of EvoLinkAI/awesome-gpt-image-2-API-and-Prompts and re-run."
  exit 2
fi

# 1. Repo identity
git remote get-url origin 2>/dev/null | grep -qE "EvoLinkAI/awesome-gpt-image-2-API-and-Prompts(\\.git)?\\b" \\
  && ok "origin remote is EvoLinkAI/awesome-gpt-image-2-API-and-Prompts" \\
  || miss "origin remote is not EvoLinkAI/awesome-gpt-image-2-API-and-Prompts (artifact may be from a fork)"

# 2. License matches what RepoPilot saw
(grep -qiE "^(CC0-1\\.0)" LICENSE 2>/dev/null \\
   || grep -qiE "\"license\"\\s*:\\s*\"CC0-1\\.0\"" package.json 2>/dev/null) \\
  && ok "license is CC0-1.0" \\
  || miss "license drift — was CC0-1.0 at generation time"

# 3. Default branch
git rev-parse --verify main >/dev/null 2>&1 \\
  && ok "default branch main exists" \\
  || miss "default branch main no longer exists"

# 4. Critical files exist
test -f "README.md" \\
  && ok "README.md" \\
  || miss "missing critical file: README.md"
test -f "cases" \\
  && ok "cases" \\
  || miss "missing critical file: cases"
test -f ".github/ISSUE_TEMPLATE/submit-prompt.yml" \\
  && ok ".github/ISSUE_TEMPLATE/submit-prompt.yml" \\
  || miss "missing critical file: .github/ISSUE_TEMPLATE/submit-prompt.yml"
test -f "contributing.md" \\
  && ok "contributing.md" \\
  || miss "missing critical file: contributing.md"
test -f "data/ingested_tweets.json" \\
  && ok "data/ingested_tweets.json" \\
  || miss "missing critical file: data/ingested_tweets.json"

# 5. Repo recency
days_since_last=$(( ( $(date +%s) - $(git log -1 --format=%at 2>/dev/null || echo 0) ) / 86400 ))
if [ "$days_since_last" -le 30 ]; then
  ok "last commit was $days_since_last days ago (artifact saw ~0d)"
else
  miss "last commit was $days_since_last days ago — artifact may be stale"
fi

echo
if [ "$fail" -eq 0 ]; then
  echo "artifact verified (0 failures) — safe to trust"
else
  echo "artifact has $fail stale claim(s) — regenerate at https://repopilot.app/r/EvoLinkAI/awesome-gpt-image-2-API-and-Prompts"
  exit 1
fi

Each check prints ok: or FAIL:. The script exits non-zero if anything failed, so it composes cleanly into agent loops (./verify.sh || regenerate-and-retry).

A curated collection of 379+ production-ready prompts and API integration patterns for GPT-Image-2 text-to-image generation, organized across 7 use-case categories (portrait, poster, UI, e-commerce, ad creative, character design, comparison) with multilingual documentation in 11 languages and real output images for each prompt. It provides developers with battle-tested prompt templates, best practices for image generation, and callable workflow skills to integrate into applications without boilerplate experimentation. Documentation-first monorepo: README.md (English source) + 10 localized variants (README_{de,es,fr,ja,ko,pt,ru,tr,zh-CN,zh-TW}.md) serve as the primary interface. Cases directory mirrors this structure with case studies (ad-creative.md, character.md, ecommerce.md, etc.) in all 11 languages. No src/ directory, no package.json—pure content repository with GitHub Issues template for prompt submissions.

AI product engineers and prompt engineers building e-commerce platforms, creative agencies, marketing teams, and SaaS applications that need production-grade image generation without spending weeks reverse-engineering effective prompts. Secondary audience: GPT-Image-2 API users exploring use cases via Evolink platform.

Early-stage but actively curated: 379+ prompts imply significant content investment, but no visible CI/CD pipeline (no .github/workflows files), no test directory, and the repo appears to be primarily documentation-driven rather than code-based. Actively maintained for content (multilingual localization exists), but lacks software engineering rigor typical of production libraries.

Low technical risk for consuming (it's documentation + prompts, not a dependency), but high risk if treating as authoritative API reference—no versioning visible, no changelog, and GPT-Image-2 API evolution could silently break examples. Single organization (EvoLinkAI) controls all content with no visible contributor guidelines or community review process. Risk of prompt drift: no automated validation that prompts still produce expected outputs.

Active maintenance of multilingual documentation and case library (all 11 language variants present). Related ecosystem projects are being developed: GPT-Image-2 × Seedance 2.0 Cinematic Workflow and GPT-Image-2 Gen Skill (callable skill library) are promoted in badges but appear to be separate repos. No visible recent commit data in provided file structure, so current velocity unclear.

git clone https://github.com/EvoLinkAI/awesome-gpt-image-2-API-and-Prompts.git && cat README.md to browse available prompts, or navigate to cases/ directory to review category-specific examples. No installation needed; this is a read-only reference unless contributing prompts via GitHub Issues (submit-prompt.yml template exists).

Daily commands: Nothing to run locally. Clone the repo and open README.md or cases/*.md files in a markdown viewer. For integration: use the referenced Evolink platform API or compatible GPT-Image-2 implementations (separate repos). No dev server, no npm/pip install.

• README.md — Primary entry point documenting the entire project's purpose, 379 curated prompts collection, and links to all translations and related tools
• cases — Core directory containing all use-case prompt files (ad-creative, character, ecommerce, portrait, poster, ui, comparison) that form the substance of this curated collection
• .github/ISSUE_TEMPLATE/submit-prompt.yml — Defines the contribution workflow for adding new prompts, establishing quality standards and submission process for community contributions
• contributing.md — Guidelines for contributors explaining how to structure prompts, follow naming conventions, and maintain consistency across the 600-file repository
• data/ingested_tweets.json — Data asset containing ingested tweet examples used to generate or validate prompt outputs
• LICENSE — CC BY 4.0 license establishing legal usage terms for all curated prompts and assets in this collection

Add a new prompt case to an existing category

1. Create a new entry in the relevant cases file following the existing format (e.g., case ID, prompt text, parameters) (cases/ad-creative.md or cases/character.md (etc.))
2. Generate example output image and save to images/case_caseXXX/output.jpg (images/case_caseXXX/output.jpg)
3. Update the localized versions of the same cases file for all 10 supported languages using consistent structure (cases/ad-creative_zh-CN.md, cases/ad-creative_ja.md, etc.)
4. Verify the badge count in README.md reflects the updated total curated prompts (README.md)

Create a new use-case category

1. Create base category file cases/newcategory.md following the structure of existing cases files (cases/newcategory.md)
2. Create localized versions for all 10 languages using the naming convention cases/newcategory_LANG.md (cases/newcategory_zh-CN.md, cases/newcategory_ja.md, cases/newcategory_es.md, etc.)
3. Add category link and description to the main README.md and all localized README files (README.md, README_zh-CN.md, README_ja.md, etc.)
4. Generate and commit example output images in appropriately named directories (images/newcategory_case1/output.jpg)

Process and submit community prompts via GitHub

1. Community member opens a new issue using the submit-prompt template which enforces structured submission (.github/ISSUE_TEMPLATE/submit-prompt.yml)
2. Maintainer reviews submission against contributing.md guidelines for quality, uniqueness, and format compliance (contributing.md)
3. Upon approval, maintainer adds prompt to appropriate cases/CATEGORY.md and all localized variants (cases/ad-creative.md, cases/ad-creative_zh-CN.md, etc.)
4. Generate output image and commit to images/ directory, update total prompt count in README badges (README.md)

Localize content to a new language

1. Create new language variants of all README files and cases files using BCP 47 language code suffix (README_pt-BR.md, cases/ad-creative_pt-BR.md, cases/character_pt-BR.md, etc.)
2. Add language badge and link to the primary README.md for discoverability (README.md)
3. Ensure all 7 case categories have translated versions following naming convention cases/CATEGORY_LANG.md (cases/character_pt-BR.md, cases/ecommerce_pt-BR.md, cases/portrait_pt-BR.md, cases/poster_pt-BR.md, cases/ui_pt-BR.md, cases/ad-creative_pt-BR.md, cases/comparison_pt-BR.md)

• Markdown (.md files) — Lightweight, version-control-friendly format for storing and versioning structured prompts; renders natively on GitHub without additional tooling
• GitHub Issues & Templates (.yml) — Enforces structured community submissions via submit-prompt.yml template; provides friction to ensure quality control before prompts enter the curated collection
• Multi-language localization (11 languages) — Maximizes reach and accessibility globally; mirrors best practices of modern open-source projects; README badges signal i18n commitment
• Git/GitHub Actions (implicit) — Enables collaborative curation, version history, and asynchronous review of prompt contributions without centralized infrastructure
• Static images (JPG) in images/ directory — Demonstrates prompt output quality without requiring live API calls; builds trust and lowers barrier to entry for users unfamiliar with GPT-Image-2

• Pure static markdown-based collection vs. dynamic web application with database

  • Why: Markdown is simpler to maintain, version-control, and host; GitHub provides free distribution; no server maintenance needed
  • Consequence: Limited ability to add dynamic features (search, filters, ratings); scaling to 1000+ prompts may require pagination or frontend migration

• Manual localization of all files vs. automated translation pipeline

  • Why: Ensures high-quality, contextually appropriate translations by human reviewers; reflects cultural nuances in prompt engineering
  • Consequence: Scaling to new languages requires volunteer translators; time-consuming to keep translations synchronized across 600 files

• Structured GitHub issue template vs. free-form contributions

  • Why: Enforces consistent metadata, clear categorization, and prompt quality standards upfront; reduces review burden
  • Consequence: Slight friction for casual contributors; may deter spontaneous submissions; requires active moderation

• Static example images vs. on-demand generation

  • Why: No API costs; instant loading; showcases real outputs for trust-building
  • Consequence: Images become stale if prompts are refined; storage costs for 600+ images; difficult to show variations

• Does not provide a runtime inference engine—collection is reference material only, users must integrate with GPT-Image-2 API themselves
• Does not handle user authentication, API key management, or account-level prompt customization
• Not a real-time collaboration platform—updates are asynchronous via GitHub pull requests and issues
• Does not include automated testing or quality validation of prompts; relies on human review
• Not designed for commercial SaaS deployment; structured as open-source educational resource

No hidden traps in traditional sense (no environment variables, no service dependencies, no database setup). However: (1) All case studies assume access to GPT-Image-2 API or Evolink platform—examples won't run without external integration. (2) Multilingual content is manually synced—no automated translation, so language variants can drift. (3) Prompts are point-in-time snapshots; no versioning, so API capability changes (e.g., model updates) can make old prompts obsolete without notice. (4) .md files lack frontmatter or metadata—poor discoverability programmatically if consuming as a dataset.

• Prompt Engineering for Vision Models — This repo's entire value proposition rests on understanding how to structure natural language instructions to guide GPT-Image-2 outputs—prompt quality directly determines image fidelity, style consistency, and commercial usability
• Zero-Shot vs. Few-Shot Prompting — The 379 prompts in this collection represent implicit few-shot examples that users can adapt; understanding when to use single-prompt (zero-shot) vs. multi-example prompts affects integration patterns for case-specific customization
• Multimodal Model Alignment (Text-to-Image) — GPT-Image-2 must align text embeddings with image latent spaces; knowing this helps explain why certain prompt phrasings (adjectives before nouns, specific style descriptors) work better than others
• Localization and Translation Strategy — This repo maintains 11 language versions synchronously; understanding content localization patterns and translation workflow is essential for scaling or contributing multilingual prompts without introducing drift
• Workflow Orchestration / Callable Skills — The companion gpt-image-2-gen-skill repo abstracts these prompts as reusable, composable units; this concept enables treating prompts as microservices or pipeline stages rather than one-off strings
• Prompt Caching and Reusability — 379+ curated prompts implicitly teach template-based prompt design; recognizing reusable prompt components (descriptors, style keywords, composition rules) enables cost-effective API usage and faster iteration
• API Design via Use-Case Taxonomy — Organizing prompts by 7 categories (portrait, poster, UI, e-commerce, ad creative, character, comparison) reveals implicit API surface design; contributors should understand this taxonomy to know where new prompts belong

• EvoLinkAI/GPT-Image-2-Seedance2-Workflow — Official companion repo providing cinematic workflow orchestration for GPT-Image-2, extending these prompts with video/sequence generation capabilities
• EvoLinkAI/gpt-image-2-gen-skill — Callable skill library that wraps GPT-Image-2 API for programmatic integration, allowing developers to use these prompts as reusable components in applications
• openai/gpt-4-vision-examples — OpenAI's official examples for multimodal image understanding; contextual reference for how GPT-Image-2 fits in broader vision-language ecosystem
• dair-ai/Prompt-Engineering-Guide — Comprehensive prompt engineering best practices and techniques that provide theoretical foundation for why these GPT-Image-2 prompts are structured the way they are
• f/awesome-chatgpt-prompts — Sister project in same 'awesome-*' tradition providing curated ChatGPT prompts; demonstrates parallel approach for text generation vs. this repo's image generation focus

To work on one of these in Claude Code or Cursor, paste: Implement the "<title>" PR idea from CLAUDE.md, working through the checklist as the task list.

Add multilingual validation and consistency check script

The repo maintains 11 translations (English + 10 localized versions) across README and all case files (cases/ directory has 55+ files). There's no automated validation to ensure all translations are complete and in sync. A new contributor could create a script to verify every case_{language}.md file exists for each base case and that README translations are all present, preventing missing translations.

• [ ] Create scripts/validate-translations.js to check all cases/{ad-creative,character,comparison,ecommerce,portrait,poster}_{de,es,fr,ja,ko,pt,ru,tr,zh-CN,zh-TW}.md files exist
• [ ] Verify all README_{language}.md files are present for supported languages
• [ ] Add GitHub Action workflow .github/workflows/validate-translations.yml to run on PRs
• [ ] Document the script in README.md with instructions for local testing

Create comprehensive CONTRIBUTING.md with submission guidelines

The repo has a submit-prompt.yml issue template but lacks a CONTRIBUTING.md file explaining the full workflow for adding new prompts, translations, and cases. With 379 curated prompts and a clear case-based structure (ad-creative, character, comparison, ecommerce, portrait, poster), new contributors need explicit guidance on file naming conventions, translation requirements, and PR submission steps.

• [ ] Create CONTRIBUTING.md documenting the case categories and file structure
• [ ] Explain naming convention: cases/{case-name}.md (English) and cases/{case-name}_{lang-code}.md (translations)
• [ ] Detail translation requirements: when submitting a prompt, which languages must be provided
• [ ] Reference the .github/ISSUE_TEMPLATE/submit-prompt.yml and explain its usage
• [ ] Add examples of a properly formatted prompt entry in each case file

Add automated prompt metadata extraction and indexing

With 379+ curated prompts across multiple case files and languages, the repository lacks a structured index or metadata system. A contributor could create a script to extract prompt metadata (title, category, language, use-case) from all case markdown files and generate a JSON index (prompts-index.json) to enable better discoverability and enable future API integrations.

• [ ] Create scripts/extract-prompts-metadata.js to parse all cases/*.md files and extract structured prompt data
• [ ] Generate prompts-index.json with fields: id, title, category (ad-creative/character/etc), language, file_path
• [ ] Add GitHub Action .github/workflows/update-prompts-index.yml to regenerate index on push to main
• [ ] Update README.md with a note about the searchable index and link to prompts-index.json

• Add automated prompt testing framework: write Python scripts in a new tests/ directory to validate that all 379 prompts in cases/ directory still produce valid images via GPT-Image-2 API, catching regressions when API behavior changes.
• Create prompt index and searchable metadata: add frontmatter (YAML) or a prompts.json index file listing all cases with tags (category, language, complexity, use-case), enabling tools to query 'give me all character design prompts for e-commerce' programmatically instead of manual markdown parsing.
• Establish community translation workflow: write CONTRIBUTING.md documenting how to translate new cases into all 11 languages (or use a translation service integration like Crowdin), since currently each new case requires manual duplication across 11 language files.

This repository appears to be a documentation/prompt collection project with minimal security risk exposure. No executable code, dependencies, credentials, or infrastructure configuration were identified. Primary concerns are limited to data governance of the ingested tweets JSON file and ensuring consistent security messaging across multiple language versions. The lack of backend code, APIs, or system configuration significantly reduces the attack surface. Recommended actions focus on data protection and documentation consistency rather than critical infrastructure vulnerabilities.

• Low · No Dependency Management File Found — Repository root. The codebase appears to be a documentation/prompt repository with no package.json, requirements.txt, or similar dependency management files provided for analysis. This limits vulnerability assessment of third-party dependencies. Fix: If this repository uses any runtime dependencies, maintain a dependency lock file and regularly scan with tools like npm audit, pip audit, or Snyk.
• Low · Potential Data Exposure via JSON File — data/ingested_tweets.json. The file 'data/ingested_tweets.json' contains ingested data that may include user-generated content or sensitive information without visible access controls or data sanitization measures. Fix: Review the contents of this JSON file to ensure no personally identifiable information (PII), credentials, or sensitive data is exposed. Implement proper data governance policies.
• Low · Multiple Language Files Increase Attack Surface — README_*.md, cases/*_*.md files. The repository contains 11 localized versions of README and case files. While beneficial for accessibility, this increases the surface area for maintaining consistency in security disclaimers and terms across all versions. Fix: Implement a centralized translation management system to ensure security notices and disclaimers are consistently applied across all language versions.

LLM-derived; treat as a starting point, not a security audit.

• Open issues — current backlog
• Recent PRs — what's actively shipping
• Source on GitHub

Generated by RepoPilot. Verdict based on maintenance signals — see the live page for receipts. Re-run on a new commit to refresh.