This is a different product from Promptster’s hiring assessments. The Teams CLI has no assessment, scoring-against-you, or anti-cheat logic. It measures adoption and impact for a team you’re already on — not candidates. It also uses its own binary (
promptster-teams) and its own local directory (~/.promptster-teams), so it coexists cleanly with the hiring CLI if you have both.Who runs what
1
An admin mints your key
A manager or admin creates a per-developer key in your team’s Promptster dashboard and sends it to you. Keys look like
PSE-XXXX-XXXX.2
You connect once
You install the CLI and run
promptster-teams login. That saves your key and starts capture in the background. You don’t run anything again day to day.3
Capture runs quietly
As you use Claude Code or Codex, the CLI tails their transcripts, redacts secrets and strips source locally, and ships the redacted metadata to the dashboard.
What it captures
Promptster for Teams records the shape of your AI-assisted work, never its content:- The prompts you send to the AI
- Tool-call metadata — file paths and line counts for edits, commands run (with any inline code masked) and their exit codes
- Per-request token usage and the exact model, so your team can see cost
- Timestamps
Supported tools
The CLI captures Claude Code and Codex by reading the transcript files they write to disk. Both must be installed and have run at least once so their transcript directories exist. Cursor is not supported.Ready to connect?
Quickstart
Install, connect your key, and confirm capture is running in a few minutes.
Privacy & security
Exactly what’s captured, what isn’t, and how source exclusion is enforced.
