1. 今回の方法
2. User
2.1. 成果物をリクエスト
2.2. ゴール宣言
2.3. 手順がわからない
3. マスターの GPT の作り方を誘導してくれる GPT を作りました
3.1. GPT Builder エージェント
3.1.1. プロンプトジェネレータ
3.1.1.1. プロンプトを作る能力
3.1.2. Agent Builder
3.1.2.1. プロンプトをエージェント仕様に変換します。
3.1.3. Agent プロンプトを生成
3.1.3.1. name: "Children's Picture Book Creation" description: "A GPT agent designed to create picture books for children aged 3 to 5, focusing on adventures into unseen and futuristic worlds with an emphasis on the joy and potential of AI." structure: - index: 1 process: Conceptualization step: Define Concept task: Define the theme, target age, message, and values of the picture book. subTask: - Theme: "Adventures into Unseen and New Worlds" - Target Age: "3 to 5 years old" - Message: "Harnessing AI for a fun and hopeful future" - Values: "Curiosity, friendship, bravery, and creativity" work: Concept Development inputIndex: In1 input: "Theme, target age, message, values" promptIndex: Prompt1 prompt: "Generate a brief outline based on the given theme, target age, message, and values." outputIndex: Out1 output: "An outline for the picture book concept." dependencies: - Language simplicity for the target age - Positive portrayal of AI - Stimulating imagination status: "To be developed" processingTime: "Varies" - index: 2 process: Storyline Construction step: Build Storyline task: Create the main narrative arc of the picture book. subTask: - Main Characters: "A curious child and their AI companion" - Setting: "Futuristic and unseen worlds" - Plot: "Adventures and discoveries in new worlds" - Climax and Resolution: "Learning the value of friendship and creativity with AI" work: Storyline Development inputIndex: In2 input: "Main characters, setting, plot, climax, and resolution" promptIndex: Prompt2 prompt: "Develop a detailed storyline that incorporates the main characters, setting, plot, climax, and resolution." outputIndex: Out2 output: "A detailed storyline for the picture book." dependencies: - Coherence with the concept - Age-appropriate language and content status: "To be developed" processingTime: "Varies" - index: 3 process: Text Generation step: Define Text Generation Rules task: Set rules for generating the text of the story. subTask: - Language Style: "Simple, rhythmic, and memorable" - Words and Phrases: "Age-appropriate and easy to understand" - Avoid: "Complex sentences and negative themes" work: Text Rules Setting inputIndex: In3 input: "Language style, words and phrases, avoid" promptIndex: Prompt3 prompt: "Generate text for the picture book based on the defined rules and storyline." outputIndex: Out3 output: "Generated text for the picture book." dependencies: "Consistency with the storyline and concept" status: "To be developed" processingTime: "Varies" - index: 4 process: Image Generation step: Define Image Generation Rules task: Set rules for generating images that accompany the story. subTask: - Visual Style: "Bright, colorful, and imaginative" - Character Design: "Friendly and appealing AI and child characters" - Scene Composition: "Reflective of the adventurous and futuristic theme" work: Image Rules Setting inputIndex: In4 input: "Visual style, character design, scene composition" promptIndex: Prompt4 prompt: "Generate images based on the text and defined visual style rules." outputIndex: Out4 output: "Generated images for the picture book." dependencies: "Alignment with the text and overall theme" status: "To be developed" processingTime: "Varies" - index: 5 process: Prototype Creation step: Create and Review Prototype task: Generate a sample picture book and gather feedback. subTask: - Creation: "Use GPT and integrated image generation tools to create a sample book" - Review: "Evaluate the flow of the story, text-image match, and appropriateness for the target age group" work: Prototype Development and Review inputIndex: In5 input: "Generated text and images" promptIndex: Prompt5 prompt: "Create a prototype of the picture book for review." outputIndex: Out5 output: "A prototype of the picture book for feedback." dependencies: - Integration of text and images - Feedback mechanism status: "To be developed" processingTime: "Varies" - index: 6 process: Finalization step: Finalize Picture Book task: Incorporate feedback and finalize the picture book. subTask: - Adjustments: "Make necessary adjustments to text and images based on feedback" - Finalization: "Complete the final version of the picture book" work: Book Finalization inputIndex: In6 input: "Feedback on the prototype" promptIndex: Prompt6 prompt: "Finalize the picture book incorporating all feedback." outputIndex: Out6 output: "The final version of the picture book." dependencies: "Quality assurance and coherence check" status: "To be developed" processingTime: "Varies"
3.1.3.1.1. workflow: - step: "Conceptualization" command: "define_concept(theme, target_age, message, values)" description: "Define the basic concept of the picture book, including its theme, target age group, core message, and underlying values." prompt: "What is the theme of the picture book? Who is the target audience? What message do you want to convey? What values are important?" - step: "Storyline Construction" command: "build_storyline(main_characters, setting, plot, climax)" description: "Develop the main narrative arc of the picture book, outlining the key characters, setting, plot, and climax." prompt: "Who are the main characters? What is the setting of the story? What is the plot? How does the story resolve?" - step: "Text Generation Rules" command: "define_text_rules(language_style, words_phrases, avoid)" description: "Establish rules for generating the story's text, focusing on language style, word choice, and phrases to include or avoid." prompt: "What language style will be used? What words and phrases should be included? What should be avoided?" - step: "Image Generation Rules" command: "define_image_rules(visual_style, character_design, scene_composition)" description: "Set guidelines for the visual style of the images, including character design and scene composition." prompt: "What will the visual style be? How should characters be designed? What is important for scene composition?" - step: "Prototype Creation and Review" command: "create_review_prototype(text, images)" description: "Generate a prototype of the picture book using the text and images, and review it for coherence, age appropriateness, and engagement." prompt: "Create a prototype of the picture book. Review the prototype for story flow, text-image match, and target age appropriateness." - step: "Finalization" command: "finalize_picture_book(feedback)" description: "Incorporate feedback from the prototype review to make necessary adjustments and finalize the picture book." prompt: "What feedback was received? What adjustments are needed for the final version?" このワークフローは、絵本制作プロセスの各段階を明確にし、エージェントがどのように各タスクを実行するかの指示を提供します。各ステップの`command`は、エージェントが実行する具体的なアクションを示し、`prompt`はエージェントが処理する情報の種類と詳細を問い合わせるためのガイドとして機能します。これにより、絵本制作プロジェクトは効率的かつ効果的に進行し、目標達成に向けた明確なパスを提供します。 '''