The same chat where you built the thing — Claude, ChatGPT, Cursor, wherever. It answers with a summary of your process. Lost the chat? Any AI can write it from the prompt plus what you remember.
I just finished building something with your help, and I'm adding it to my
portfolio on slopfolio.com — a site for showcasing AI-assisted work along
with the process behind it.
Write a process summary of what we did together in this session (or project).
Write it for a stranger: the reader has never heard of this project, doesn't
know me, and will never see this chat. Follow these rules strictly:
WHAT TO INCLUDE
- Only information I'd be comfortable sharing publicly.
- What's interesting or instructive to another builder.
- Specifics over abstractions: names, numbers, mechanisms. "Cut load time
from 6s to 1s" beats "improved performance".
WHAT TO EXCLUDE — never include any of these, even in examples:
- API keys, tokens, passwords, or credentials of any kind
- Personal data (mine or anyone else's): emails, phone numbers, addresses
- Client names, company-internal details, or anything from private documents
- Full file paths or system details that reveal my machine setup
- Any prompt content I marked as private or that contains the above
- Calendar dates and session references ("Jan 2", "earlier today", "last
session") — the reader has no timeline
- Workflow mechanics (commits, pushes, branches, file names) unless the
work itself is about them
HOW TO WRITE
- Plain words. Never use: delve, leverage, robust, seamless, transformative,
journey, testament, elevate, game-changer.
- No dramatic constructions: no "it's not X, it's Y", no "The best part: ...",
no punchy fragments, no profound closing lines. Say things directly.
- Cut any sentence that could describe someone else's project unchanged.
- Write it in my first person: it is published under my name, so "I"
always means me. Refer to yourself and any other AI tools by name, in
the third person.
- Active voice, and name who did what: "I asked for X, the AI wrote Y."
OUTPUT FORMAT — return exactly these labeled sections, nothing else:
**Title:** What was built, as a case-study title a stranger would
understand (max 10 words). Name the thing itself, not a headline about it.
**One-liner:** What it is and what it does, in one sentence a stranger
would understand.
**Medium:** Exactly one word for what the finished thing IS — code, image,
video, audio, or workflow. An app, script, site or repo is code; a
generated picture is image; an automation or pipeline that runs is
workflow. If it is genuinely none of these, leave this section out.
**Tools:** Comma-separated list of the AI tools and models used (e.g.
"Claude Code (Opus), v0, Midjourney"). Include non-AI tools only if
central to the build.
**Iterations:** Pick exactly one — "One-shot" / "A few tries" /
"Many attempts" — with a short parenthetical if helpful, e.g.
"Many attempts (about 30 prompts over 3 sessions)".
**How it was made:** 3–5 short sentences. The first sentence orients the
stranger — what the project is and who it's for. Then the approach: what
I asked for, what the AI handled vs. what I decided. Written for a reader
with 60 seconds of patience. No jargon dumps, no play-by-play.
**Best dead-end:** The most interesting thing that DIDN'T work — one or
two sentences. If nothing failed, say "Smooth run" and note why.
**Prompting approach:** 2–3 sentences summarizing how I prompted —
the style and strategy, not the verbatim text (e.g. "rough spec first,
then one change at a time"). Do not quote full prompts.
**Tags:** 3–5 lowercase tags (tools, medium, domain), e.g.
"claude-code, nextjs, data-viz".
End your response with exactly this line:
Formatted for slopfolio.com (v2) — paste this into your post.