Which archetype are you?
The Architect
A name for the shape your six axes make, and how rare that shape is. The only part you can post without explaining it first.
Sample report · six axes
Hover an axis
Sample data. Your report is computed from your own transcripts, run the command below.
What it finds
Things your transcripts already know about you.
No questionnaire and no self-report — each finding is computed from something your sessions already recorded. Hover any card.
How many agents do you run?
6 agents in parallel
Counted from parallel tool calls and subagent runs — how much you fan out, and whether you delegate the branches or drive each one yourself.
How often do you plan?
62% in plan mode
Whether you make the agent commit to a plan before it writes, and how often you open in plan mode first.
What do you push back on?
Taste, measured in churn
Lines the agent added and you had it take out again inside the same session — what you insist on once the code exists.
How do you steer?
One-word corrections
Read from turns that land mid-tool-use. A one-word course correction is a different instrument from a rewritten paragraph.
What does your AI think of you?
A second-person read
What a model infers about how you hand over work, from the way you actually hand it over — with session ids attached to every claim.
How much ground per session?
31 files, one sitting
Files touched, lines moved, and how long you stay in a session before starting a new one.
Does your first plan survive?
Two revisions in
How often the first plan comes back rejected before you let the agent write. A rejected plan is a labelled objection, not a false start.
What do you pin down first?
Spec before token one
How much you pin down before the first token, measured against how much you had to add afterwards.
Getting the report
One command, and it reads what is already there.
01
It reads what is already there
Claude Code, Codex, opencode, Cursor. Nothing to install or turn on.
02
Redacted before it leaves
Credentials and your home path go first. The upload aborts if anything slips.
03
Your link, your email
Claim it with one click. Needs 5 sessions or more to score.
Upload · one command, then sign in for the report
Two ways to upload coding sessions.
It looks at the AI session transcripts on your computer. Where you run the command tells it whether to analyze every project at once or just one.
Already ran the command? Sign in to see your report.
Option 1 · recommended
All my repos
Best for a broader picture across projects. Change into the parent folder that holds your repos, then run.
curl -fsSL https://prduck.tryproduck.com/upload.sh | bash
Option 2
Just one repo
Best for focusing on a single project. Change into that project’s folder (replace ~/path/to/your-project with the real path) and run.
cd ~/path/to/your-project && curl -fsSL https://prduck.tryproduck.com/upload.sh | bashOr let an agent find them
You can use this prompt in Claude, Codex, or Cursor to find all the repos on your machine with agent transcripts, show you the list, and hand back ready-to-run commands for the ones you pick.
Find every repo on my machine where I've used a coding agent. Check ~/.claude/projects, ~/.codex/sessions, ~/.local/share/opencode/storage, ~/.cursor/chats. For each repo, list name, absolute path, and total session count. Ask me which ones to include. For each one I pick, hand back this command with the path filled in: cd <ABSOLUTE_PATH> && curl -fsSL https://prduck.tryproduck.com/upload.sh | bash
The command reads agent transcripts under ~/.claude/projects, ~/.codex/sessions, ~/.local/share/opencode/storage, ~/.cursor/chats, strips credentials and your home directory on your machine, aborts if anything slips through, then uploads the redacted transcripts so the server can score them. Full prompt, plan and diff text is included in the upload. How that data is handled.
Nominate
