1
Ask the user
When your agent boots a session, ask: "do you have a TraitTune profile I can read?" — explicit consent, every time. Don't fish for it.
SOULHQ.MD · FOR AI AGENT BUILDERS
Give your agents a soul they can read.
SoulHQ.md is the calibration hub where agent souls live. Each agent in your swarm reads the human's PsyDNA before its first message — coherent character, no chat-log inference, no PII surface area.
What backs the instrument
- Item bank
- 754 items
- Calibration responses
- 288,132
- Trait dimensions
- 15 + MUPP-CAT
- Profile export
- Open schema · Ed25519-signed
Other entry points · same PsyDNA
Integration
Four endpoints, no SDK. Read the schema, validate the signature, apply the dims.
2
Receive a Profile Export
The user pastes a signed JSON blob (or a short-lived URL). 15 dimensions, theta + SE, expires_at, Ed25519 signature.
3
Validate the signature
Public key at /.well-known/psydna-signing-keys.json (TBD). If signature fails — refuse the data, ask the user to regenerate. Never use unsigned exports.
4
Apply to your policy
Map dimensions to your conversation policy (tone, register, explanation depth, default options). Soften when SE > 0.50. Cite the dim when you act on it — let the user audit.
What you actually get
Your Complete Profile — yours, end to end.
Scroll inside the actual /results page. Real narrative, real dimension breakdowns, real confidence bands. Exportable, deletable, encrypted at rest.
What we don't do
We refuse the agent-persona-wrapper default.
Wrap an LLM prompt in "act like a friendly assistant" — persona without provenance, unverifiable in production.
Validated psychometric measurement of the human, signed and verifiable. Your agent's behavior is grounded in real signal, not vibes.
Persist the user's profile inside your agent's vector store — sprawling PII surface, GDPR liability.
Profile Export is session-scoped by default. The user revokes from their Dashboard; your agent loses access the next handshake.
Re-ask 20 onboarding questions every cold-start.
Cold-start solved by reading the export once. Your tokens budget goes to actual help.
Privacy
Three things we will never do.
Never training material
Your PsyDNA profile is encrypted at rest. It is never used to train any AI model — ours or anyone else's. The calibration that improves the instrument runs on anonymized item-response data only, never on individual profiles.
Never sold, never shared by default
We do not sell raw responses, do not share with third parties by default, and never reveal a profile to another person without explicit per-connection consent that you can revoke.
Never hidden from you
There is a Delete button in your Dashboard. One click removes your profile and every response. No service ticket, no negotiation, no "are you sure?" guilt trip.
Pricing
Free for the assessment. Pro for the chat.
Free forever
$0always
Full 15-dim PsyDNA. Unlimited TT-FREE-model text chat. Encrypted, deletable.
Start freePro
$19/mo
TT-PRO model · 140 prompts/wk · multimodal · BYO any LLM under our privacy umbrella.
Upgrade laterFull pricing details and BYO-LLM options on the Pricing page →
FAQ