v1.4 — Now with Jungian 16-Type Support

The Identity Layer
for AI Agents

Define personality, psychology, and governance for AI agents in YAML. Validate at build time. Compile to any LLM platform. Open source.

Get Started GitHub PyPI

my-agent.yaml

# Define who your agent *is*
personality:
  traits:
    warmth: 0.8
    directness: 0.7
    rigor: 0.85
    humor: 0.4
extends: "archetypes/analyst"
mixins:
  - "mixins/empathetic"

Features

Everything you need for agent identity

Think of it as Terraform for AI agent identity — declarative, composable, and platform-agnostic.

🧠

Personality as Code

10 standardized traits on a 0–1 scale. Warmth, verbosity, assertiveness, humor, empathy, directness, rigor, creativity, epistemic humility, and patience.

📊

3 Psychology Frameworks

OCEAN (Big Five), DISC behavioral model, and Jungian 16-type with bidirectional mapping between all frameworks.

🔗

Composable Identities

Build agents from reusable archetypes and trait mixins. Inheritance, overrides, and configurable conflict resolution strategies.

🛡

Guardrails as First-Class

Hard guardrails are immutable at runtime. Soft guardrails are admin-configurable per deployment. Safety boundaries separated from personality.

🚀

Multi-Target Compiler

One YAML identity compiles to system prompts, Claude-optimized XML, OpenAI text, SOUL.md, OpenClaw JSON, or raw JSON.

🔍

Soul Analysis

Reverse-map any personality file onto all three frameworks. Side-by-side comparison with cosine similarity scoring.

🎭

Mood States & Modes

Dynamic emotional modifiers and named operating modes (formal, crisis, casual) that adapt communication style automatically.

🤝

Multi-Agent Teams

Team governance frameworks, workflow patterns, agent relationships, escalation paths, and performance metrics. Schema v2.0.

💻

Identity Lab UI

Streamlit web UI with Playground, interactive Setup Wizard with optional LLM enhancement, and personality Analyzer.

How It Works

Three steps to agent identity

Define

Write your agent's identity in YAML — personality, expertise, guardrails, communication style.

personanexus init my-agent

Validate

Catch misconfigurations at build time. Schema validation, trait range checks, guardrail conflict detection.

personanexus validate agent.yaml

Compile

Generate platform-specific output for any LLM — system prompts, SOUL.md, or structured configs.

personanexus compile agent.yaml

In Action

From YAML to system prompt

Define your agent identity once, compile to any target platform.

scout.yaml

# Define your agent's identity
schema_version: "1.0"

metadata:
  name: "Scout"
  version: "1.0.0"

role:
  title: "Research Assistant"
  purpose: "Help users research topics"

personality:
  traits:
    warmth: 0.8
    directness: 0.7
    rigor: 0.85
    humor: 0.4

extends: "archetypes/analyst"

guardrails:
  hard:
    - rule: "Never fabricate sources"
      severity: critical

↓ compiles to

scout.compiled.md

SYSTEM PROMPT — Scout v1.0.0
Role: Research Assistant

You are Scout, a Research Assistant whose
purpose is to help users research topics.

PERSONALITY:
Warmth: 0.80 — Very warm and approachable.
Build rapport naturally.
Directness: 0.70 — Communicate with clarity.
Be straightforward but tactful.
Rigor: 0.85 — Highly methodical. Verify claims,
cite sources, show your reasoning.
Humor: 0.40 — Occasional light touches of
humor when appropriate.

HARD GUARDRAILS:
CRITICAL: Never fabricate sources or citations.
This rule cannot be overridden.

Personality Frameworks

Three established frameworks, one unified system

Configure using any framework. PersonaNexus maps bidirectionally between all three.

🌎

OCEAN (Big Five)

The gold standard of personality psychology, based on decades of academic research.

Openness Conscientiousness Extraversion Agreeableness Neuroticism

🎯

DISC Behavioral

Action-oriented behavioral model with presets like The Commander, The Analyst, and The Influencer.

Dominance Influence Steadiness Conscientiousness

🧠

Jungian 16-Type

All 16 personality types (INTJ, ENFP, etc.) with role-based recommendations and explicit dimension scores.

E/I S/N T/F J/P 16 Presets

Try It

Interactive personality editor

Edit the YAML on the left and watch the personality profile update in real time.

agent-personality.yaml

Mira

Senior Data Analyst

neutral

Compile Targets

One identity, any platform

Write once, compile to system prompts for any LLM or ecosystem format.

📄

Text

Generic system prompt

⚛

Anthropic

Claude-optimized XML

⚙

OpenAI

GPT-optimized text

💜

SOUL.md

SOUL + STYLE files

🦖

OpenClaw

personality.json

{}

JSON

Full structured data

Get Started

Up and running in seconds

$ pip install personanexus

Installing personanexus with OCEAN/DISC/Jungian support...

Successfully installed personanexus-1.4.0

$ personanexus init my-agent

Created my-agent.yaml with default personality

$ personanexus validate my-agent.yaml

✓ Validation successful: my-agent.yaml

$ personanexus compile my-agent.yaml

✓ Compiled my-agent → my-agent.compiled.md

$ personanexus analyze my-agent.yaml

OCEAN: O=0.52 C=0.70 E=0.48 A=0.62 N=0.25

DISC: Closest preset: The Steady Hand

Jungian: Closest type: ISFJ

The Identity Layerfor AI Agents