Your project, ready to share

One work per member. Review your submission here, or update it and submit a new version.

FROM YOUR PROJECT TO YOUR STORY

Prepare your project for sharing

Give one instruction to your Agent or development AI. Turn what you built and how you use it into a diagram and a story others can understand.

Choose what you actually built
Explain what your Agent helps you do and which L2 ideas you applied. You do not need to implement every lesson or feature; focus on what you need and want to share.
Let your AI help with the image and story
Copy the instruction below to the AI that knows your project. It will ask about missing context, then prepare a flowchart, your experience, and an Agent-perspective introduction.
Check the content and submit
Check facts and image text. Blur or cover sensitive details and remove passwords, API keys and tokens. Add a title, paste the description and upload 1–3 images. Never present unbuilt features as results.

Your AI cannot generate images? The same instruction asks it for a complete image prompt. Paste that into ChatGPT or another image-capable AI to create your picture.

Traditional Chinese example: requests connect to Agent capabilities, a second brain and a feedback cycle. — Layout reference in Traditional Chinese. Your AI should use your actual features and leave out this author’s signature.

Add a real usage example (optional)Included in the copied instruction only. Use a short example you are comfortable sharing.

Read or manually copy the full instruction

Help me prepare a public project showcase for my own AI Agent: a user-experience flowchart and copy I can paste into AI Agent School.

First, understand my actual Agent
- If you are my Agent, use the capabilities and usage context you are already authorized to access. If you are my development AI, inspect the current project and its documentation within your existing permissions; do not assume that implemented features have actually been used.
- Distinguish implemented capabilities, real usage confirmed by me or available evidence, and future ideas. Never invent usage history, results, time saved, integrations, memory, or self-improvement.
- If important information is missing, ask at most three short questions together: what I actually ask it to do, one real example and its outcome, and how it retains knowledge or incorporates feedback. Wait for my answers before claiming actual usage. Do not demand features I have not built.
- Use only content suitable for public sharing. Summarize and anonymize; do not export private chats, personal or client details, credentials, file paths, or internal instructions. Ask only for a short sanitized example when needed.

Deliver these three items in English:

1. A flowchart people can understand at a glance
- Describe the user experience, not software architecture. Start with how I ask for help, connect to my Agent, then show 3–6 verified capability groups and the useful results I receive. Use fewer groups if that is all I have built.
- Use short plain-language labels, simple icons, clear arrows, ample spacing, and a readable portrait layout (4:5 preferred). Use a concrete, benefit-led title based on my project, not a generic school-branded Agent name. No persona names, technical data flows, APIs, databases, code, decorative corner text, or invented features.
- If supported, show how real material becomes reusable knowledge (my second brain) and how my feedback improves future help. Connect those steps back to the next interaction. Describe the real controls: for example, changes requiring my confirmation. Do not imply autonomous training, perfect memory, or unlimited growth.
- If those abilities are absent, omit them. Keep future plans out of the current-capability diagram. Never copy all capabilities from a reference image.
- Choose a clear visual style suited to my project; match the diagram with the descriptions below. Do not add a signature, personal name, school promotion, URL, or QR code unless I explicitly request it. A supplied reference image is a layout reference only, not evidence of my features or permission to copy its signature.
- If you can generate images, use your image tool and return the actual downloadable image. Aim for a static PNG/JPG/WebP under 5 MB. Verify labels and arrows if you can inspect the result; do not claim unperformed checks.
- If you cannot generate an image, or generation fails, return a complete standalone image-generation prompt in a clearly labeled code block for ChatGPT or another image-capable AI. Include the exact final title and every label, node, arrow, group, feedback loop, layout, colors, and legibility requirements. Resolve the content using my actual project first; no placeholders, “see above”, dependency on private files, or request for the receiving AI to invent functions. Explicitly ask it to generate the image. Mermaid, ASCII, or an SVG/code listing alone is not the requested shareable image; you may provide them as optional extras only. Never claim an image exists when it does not.

2. Work title and ready-to-paste description
- Work title: at most 100 characters.
- One description, at most 4,000 characters total (including both voices, headings, spaces and line breaks), preferably 1,000–2,000 characters. Use two short sections:
  “How I use it”: write in my first person. Explain why I built it, what I ask it to do, one confirmed example from request to result, and the L2 learning I applied if known. Do not invent completed lessons.
  “How my Agent helps me”: write in the Agent's first person, identifying this as an Agent-perspective draft. Explain how I usually ask for help, what it does for me, where I check or confirm the result, and how knowledge and feedback help next time if actually supported. The voice may be personal but must not claim hidden access, consciousness, or independent learning it does not have.
- Separate verified current abilities from optional next steps. If there is no confirmed usage yet, say it is an implemented capability awaiting real use, not a personal success story. Keep the writing natural and concrete, without marketing hype or unsupported numbers.

3. A short handoff
- Explain which text goes into “Work title” and “Work description”, and which image to upload. If image generation is unavailable, tell me to paste the supplied image prompt into ChatGPT or another image-capable AI and save its output first.
- Remind me to check accuracy, privacy and image text before sharing. The form accepts 1–3 static JPG/PNG/WebP images, up to 5 MB each and 15 MB total. If exporting or compressing is needed, say so honestly.
- I will review and submit myself; do not publish, submit, upload to the school, or change my Agent's settings on my behalf.