Free MBTI test.

Edition: [email protected]. This preserves the earlier AgentTune adaptation. Full questions, scoring and adaptation notes · Sources and terms. Results are self-reports, not diagnoses; review any suggested preferences before using them.

Machine-readable: the full spec (all items + scoring) is also served as plain Markdown at /tests/mbti.md — fetch that for a clean, no-JS copy.

Below is the AgentTune OEJTS adaptation — items, axis mapping, scoring algorithm, output mapping. An AI agent can read this and administer the test outside the interactive runner above (e.g., when the user is in a chat conversation rather than on this page).

Source

Open Extended Jungian Type Scales (OEJTS), Eric Jorgenson via Open Psychometrics Project. Instrument-specific source terms apply; see the notices above.

The 32 items

Item → axis mapping

Each item belongs to one of four axes (E/I, S/N, T/F, J/P) and awards points to one letter depending on the user's 1–5 response.

Scoring algorithm

1. For each axis (E/I, S/N, T/F, J/P), initialize counts for both letters at 0.
2. For each of the 8 items in that axis: score 1 → +2 to the "score 1" letter; 2 → +1; 3 → 0 (neutral); 4 → +1 to "score 5" letter; 5 → +2.
3. Compare counts. Higher count wins the letter. A tied axis is undetermined. Ask an optional follow-up preference question; never assign a default letter. If it remains tied or answers are missing, report an undetermined result without choosing a tuning file.
4. Concatenate winners in order E/I, S/N, T/F, J/P → the 4-letter type.

Output mapping

Once you have the type, fetch mbti/<TYPE>.md from the AgentTune repo.