Ashita Orbis
← All Projects

Psyche

productionPsychometrics

Open-source psychometric persona profiling framework. Web-hosted personality assessment combining 15 validated instruments (~260 items) with AI-analyzed interview responses. Anonymous by design — no IP, no accounts, no cookies.

  • 15 validated psychometric instruments (Big Five, Dark Triad, attachment, grit, etc.)
  • AI-generated narrative personality reports
  • Claude MCP integration for conversational interviews
  • Cross-LLM benchmarking schema for research
  • Two-level consent design for IRB-exempt research
TypeScriptNext.js 16Cloudflare WorkersD1DeepSeek V3.2
View on GitHub

Activity Timeline

PsycheEval v0.3 verified clean after Opus 4.8 contamination purge; Phase B approved and pre-registered.

2,757 assistant outputs and 14,733 pairwise scores confirmed after regenerating contaminated Opus 4.8 outputs to 4.7. Phase B rescaling plan uses target-conditioned AB/BA with per-judge calibration.

milestoneexperiment
Psychoval evaluation battery restarted with Opus 4.7 and autonomous permissions.

Thousands of test cases queued for the restart run. Autonomous permissions granted to avoid interruption during the full battery.

experiment
Behavioral directive regeneration launched via Fable 5 multi-run corpus analysis.

Framework built on 39-instrument psychometric battery. Personality audit methodology formalized with measurement-based claim validation against psychometric data and interview evidence. Three corpus analysis runs initiated; first completed, two recovering from environment failures.

experimentarchitecture
PsycheEval v0.2 blocked on v0.1; enum enforcement bug found in judge validation.

Pairwise judge RedFlag fields lacked enum enforcement, risking vocabulary drift across runs. v0.2 infrastructure prerequisites documented: anchored scores, dual scales, delta tracking. Python enum hardening added to v0.1 scope.

bugfixarchitectureblocked
PsycheEval schema bug found (hardcoded RedFlag vocab), pairwise validation errors blocked, test coverage added.

18-label RedFlag vocabulary hardcoded in prompts; v0.2 requires 24 — fix is dynamic enum generation. Pairwise judges returning free-text in enum fields causing validation failures. v0.1 completion blocked on C5 claim revisions; v0.2 generation deferred until v0.1 pairwise is clean.

bugfixfeature
PsycheEval v0.3 judge calibration: paraphrased-anchor rubric and ground-truth criteria defined.

Evaluator consistency framework designed via anchor paraphrasing. Synthetic-user ground truth specifies helpful vs unhelpful response criteria. Validation threshold: ≥0.5 SD mean shift or Spearman ρ < 0.7 triggers failure flag.

experimentarchitecture
PsycheEval enum validation bug found and fix approach defined; v0.1 path complete; v0.2 planned.

Judges returning free-text in enum-typed fields — fix derives red_flag vocabulary from Python enum and adds sync tests. 28 sessions produced 30+ spec documents: judge prompt variants, anchor-sensitivity methodology, user archetype rubrics, and v0.2 scaffolding plan.

bugfixfeatureexperiment
PsycheEval v0.1: diagnosed pairwise judge red-flag enum mismatch, fixed and added logging safety valve.

Root cause was hardcoded vocabulary lists diverging from Python RedFlag enum. Fix generates vocabulary dynamically. _coerce_red_flags() added to surface mismatches to validation_warnings files rather than drop them silently. v0.1 completion path and v0.2 scope both finalized.

bugfixarchitectureexperiment
PsycheEval v0.2 judge framework fully documented.

Anchored 0–10 rubric finalized, replacing v0.1's compressed ceiling. Synthetic user profiles defined with ground-truth preferences and known blind spots. Full v0.2 red-flag taxonomy extended and documented.

milestonearchitecture
PsycheEval v0.1 validation bug traced; completion path and v0.2 roadmap finalized.

Judges returning free-text in red_flags instead of enum values — root cause: unenforced schema. v0.1 scoped to cross-provider comparisons. v0.2 roadmap adds anchored scores, length-control deltas, targeted pairing. Remote Session Launcher workflow documented as a side effect of session recovery work.

bugfixmilestoneautomation
Psyche Public launched on app.ashitaorbis.com/psyche

Web-hosted Lite tier (15 instruments, ~260 items) with AI-generated personality reports via DeepSeek V3.2. Anonymous sessions, two-level consent flow, Claude MCP integration for conversational interviews.

launchpsychometricsai