Skip to main content
For API access, use your API token (obtained from the dashboard under Settings → API Keys) in the Authorization: Bearer header. See Authentication for details.
1

Sign in to Promptster

Sign in at promptster.ai using your organization account. You can sign in via SSO or email.To get your API token, go to Settings → API Keys in the dashboard and copy your token.
2

Create an assessment

Post your task definition. The taskBrief describes what the candidate should build.
The response includes the new assessment’s id. Save it — you’ll need it to generate keys and retrieve sessions.
You can generate up to 50 candidate keys per assessment. Set expiresInDays to control how long a key remains valid after generation — after expiry, the key cannot be redeemed.
3

Generate candidate keys

Generate one key per candidate. If you include candidateEmails, Promptster sends invite emails automatically.
The response contains the generated keys. Each key has the format PST-XXXX-XXXX.
candidateEmails and candidateNames are optional and positional — the first email is paired with the first name. If you omit them, keys are generated without any candidate association and no email is sent.
4

Candidates complete the assessment

Share the PST-XXXX-XXXX key with each candidate. They install the Promptster CLI and run:
This configures their AI tool — Claude Code, Codex, or Cursor — to capture session telemetry automatically. The candidate works normally and runs promptster done to submit when finished.
5

List sessions for your assessment

Once a candidate submits, their session appears in the list.
The response is a paginated list of sessions. Each session includes summary fields like eventCount, decisionFlagCount, and aiAttributionPct.
6

Review the session timeline

Fetch the full chronological event log for a session. The timeline returns events in order, paginated by limit (max 2000 per request).
Use the after cursor for subsequent pages:
The response includes a hasMore flag indicating whether additional pages exist.
7

Review decisions

Fetch the architecture decisions the candidate captured during the session.
Each decision is a decision_event timeline row with the candidate’s choice, rationale, and tradeoffs.

Next steps

Session review

Understand the full session data model and how to interpret signals.

Cohort stats

Compare a candidate’s metrics against the rest of the cohort.

Manage assessments

Full reference for assessment and key management.

API reference

Complete API reference for all reviewer endpoints.