Concepts & events

Many classrooms allow AI tools. Showtail is built around a simple idea: using AI is easier to evaluate when students can show their process.

What Showtail is

• A show your work tool for coursework and projects.
• A hands-free way to document your process: your prompts and edits are captured automatically as you work with AI.
• A local, file-based trail that is easy to commit to git and easy for a human to review.
• Evidence of your process, showing how you worked with AI to build the project.
• Team-aware. On a group project, each student gets their own folder under one shared .showtail/, so trails merge through git without conflicts and the report can be per-student or a combined team view.

What Showtail is not

• Not an AI detector. Showtail does not try to guess whether work was AI-generated.
• Not surveillance. It only records your prompts and file edits while a capture integration is enabled — nothing else.
• Not a cloud service. Nothing leaves your machine, and there are no external API calls.
• Not a grading tool. It produces a report so people can review the work. It does not judge the work itself.

Event types

Showtail records four kinds of event, all captured automatically while you work:

Type	What it is
prompt	A prompt you sent to an AI tool
ai_output	An AI response
artifact	A file you created or changed
decision	A choice you made when the AI paused to ask you to pick between options

Every event records:

• id
• ISO timestamp
• type
• text
• optional files
• the tool it came through (e.g. claude-code, github-copilot, codex, chatgpt, google-gemini)
• optional tags
• the current git commit hash, if your project is a git repository
• actor: "student"

Captured text is scrubbed for secrets and personal data before it is stored (see Privacy & redaction).

Where it goes

Each event is appended to a local, file-based trail under .showtail/. See Data layout for the exact structure, and Example report for what the rendered output looks like.