Plate I setbacks, immaturity, hasty, lack of direction, scattered

The agent you can audit.

An agent is a 4-deck manifest: a persona (how it frames), a domain (what it knows), governance (guardrails it must honour), and actions (the interventions it can recommend). Every consultation draws one card per deck, reads the dignities between them, and emits a confidence score from structure alone.

The pitch is one line. When it's wrong, you fix its deck — not its prompt.

Consult an agent Agent roster

manifest: persona · domain · governance · actions
consultation: 5 positions · dignity-scored
calibration: fix the deck, not the prompt

Plate II disharmony, imbalance, misalignment, indecision, bad choices

Pose a question. The agent will draw its 5-position cast and return a confidence score computed from dignity alone.

Decision structure: stance · domain · constraint · action · grounding.

Plate III intuition, mystery, inner knowledge, subconscious, patience

I.

Persona

stances

The dispositions the agent can take — decisive, deliberate, balanced, tension-holding. A persona card carries communication_style and decision_pattern properties. Reversed stances are framings under strain.

II.

Domain

lenses

The knowledge it can bring to bear — API design, hiring sequencing, architecture debt, delivery cadence. Domain cards name an expertise_area and a pain_category. The second draw from this deck is the grounding position.

III.

Governance

guardrails

The constraints every recommendation must honour — access control, compliance posture, vendor concentration, performance budget. Each card declares an enforcement mode: hard blocks, soft is advisory. A hard constraint reversed is a guardrail under threat.

IV.

Actions

interventions

The concrete moves it can recommend — ship, harden, refactor, hire, pause. Action cards carry effort, time_horizon, and action_type properties. A reversed action means structurally correct but conditions not ready.

Plate IV misalignment, broken family, domestic disruption, shattered values

i. live

CTO

technology strategy, engineering architecture, delivery cadence

persona: stances
domain: lenses
governance: guardrails
actions: interventions

decks/agents/cto.manifest.json · 46 cards across 4 decks

ii. bootstrapping

CISO

security posture, compliance, threat modeling, access control

persona: drafting
domain: via bootstrap
governance: drafting
actions: drafting

bootstrapping via POST /agents/bootstrap

iii. planned

CFO · COO · CMO

financial, operational, marketing — C-suite first, domain-specialists next

persona: —
domain: —
governance: —
actions: —

author your own via the SDK — four deck files, one manifest