---
title: Traits
description: Define who an agent is — personality dimensions that influence behavior.
---

A **Trait** defines a personality dimension that influences HOW an agent works. Traits are declarative — you don't tell the agent what to do, you tell it who it is. The system assembles runtime parameters from trait declarations.

## Structure

```go
type Trait struct {
    cortex.Entity
    ID          id.TraitID
    Name        string
    Description string
    AppID       string
    Dimensions  []Dimension
    Influences  []Influence
    Category    TraitCategory
    Metadata    map[string]any
}
```

## Dimensions

Each trait has one or more bipolar dimensions — axes with labeled extremes and a value from 0.0 to 1.0:

```go
type Dimension struct {
    Name      string  // dimension name
    LowLabel  string  // label for 0.0 end
    HighLabel string  // label for 1.0 end
    Value     float64 // position on the axis (0.0–1.0)
}
```

Example: a "thoroughness" trait with a "depth" dimension:

```go
Dimension{
    Name:      "depth",
    LowLabel:  "surface-level",
    HighLabel: "exhaustive",
    Value:     0.85, // leans heavily toward exhaustive
}
```

## Influences

Influences define how a trait modifies agent behavior at runtime:

```go
type Influence struct {
    Target    InfluenceTarget // what to modify
    Value     any             // the modification value
    Condition string          // optional condition
    Weight    float64         // influence strength (0.0–1.0)
}
```

### Influence targets

| Target | Effect |
|--------|--------|
| `prompt_injection` | Injects text into the system prompt |
| `temperature` | Modifies the LLM sampling temperature |
| `max_steps` | Adjusts the maximum reasoning steps |
| `tool_selection` | Influences which tools are preferred |
| `response_style` | Modifies output formatting |

## Categories

Traits are organized into four categories:

| Category | Description |
|----------|-------------|
| `personality` | Core personality dimensions (e.g., openness, conscientiousness) |
| `workstyle` | How the agent approaches tasks (e.g., thoroughness, speed) |
| `communication` | Communication preferences (e.g., directness, empathy) |
| `risk` | Risk tolerance and decision-making style |

## Store interface

```go
type Store interface {
    CreateTrait(ctx context.Context, trait *Trait) error
    GetTrait(ctx context.Context, traitID id.TraitID) (*Trait, error)
    GetTraitByName(ctx context.Context, appID, name string) (*Trait, error)
    UpdateTrait(ctx context.Context, trait *Trait) error
    DeleteTrait(ctx context.Context, traitID id.TraitID) error
    ListTraits(ctx context.Context, filter *ListFilter) ([]*Trait, error)
}
```

## API routes

| Method | Path | Description |
|--------|------|-------------|
| `POST` | `/cortex/traits` | Create a trait |
| `GET` | `/cortex/traits` | List traits |
| `GET` | `/cortex/traits/{name}` | Get a trait by name |
| `PUT` | `/cortex/traits/{name}` | Update a trait |
| `DELETE` | `/cortex/traits/{id}` | Delete a trait |