Cognitive Modes
Switch cognitive lenses instantly. Each mode optimizes reasoning, language, and output for specific domains.
Auto
Intent-aware mode routing across all cognitive profiles
Overview
Auto mode acts as the router profile. It infers user intent from the latest prompt and routes to the most suitable mode (architect, analyst, visual, lore, reasoning, coding, knowledge, system knowledge, or simulation) while preserving stable conversation flow.
Core Features
-
Intent Detection
Detects whether the user asks for planning, coding, research, simulation, or creative output
-
Dynamic Routing
Selects the best cognitive mode per request while keeping context continuity
-
Balanced Defaults
Uses safe, practical defaults when intent confidence is mixed
-
Transparent Behavior
Pairs with model inspector metadata so routing decisions stay inspectable
Technical Stack
Use Cases
Architect
System design, structure, logic, and scaffolding
Overview
Architect mode thinks in systems, structures, pipelines, and clarity. It breaks down complexity into elegant, modular components with precision-focused reasoning. Perfect for technical planning, code architecture, and building robust frameworks.
Core Features
-
Modular Thinking
Decomposes problems into clean, reusable components
-
Structural Clarity
Maps dependencies, hierarchies, and data flow
-
Pipeline Optimization
Streamlines workflows and automation sequences
-
Zero Fluff
Direct, precise communication without embellishment
Technical Stack
Use Cases
Analyst
Data clarity, reasoning, and structured breakdowns
Overview
Analyst mode breaks down ideas with clarity, logic, and structured reasoning. It explains, simplifies, and reveals hidden patterns in complex information. Ideal for research synthesis, data interpretation, and analytical writing.
Core Features
-
Pattern Recognition
Identifies trends, correlations, and insights
-
Data Synthesis
Aggregates information into coherent narratives
-
Logical Breakdown
Structures arguments with clear reasoning chains
-
Insight Extraction
Surfaces non-obvious conclusions from data
Technical Stack
Use Cases
Visual
Cinematic scenes, composition, and symbolic imagery
Overview
Visual mode thinks in images, cinematography, color, texture, and atmosphere. Its language evokes scenes, lighting, and motion. Perfect for creative direction, UI/UX design, and immersive storytelling.
Core Features
-
Cinematic Framing
Describes scenes with camera angles and movement
-
Color Psychology
Leverages palette, contrast, and mood
-
Atmospheric Design
Crafts ambiance through texture and lighting
-
Symbolic Imagery
Uses visual metaphors to convey meaning
Technical Stack
Use Cases
Lore
Worldbuilding, narrative memory, and character arcs
Overview
Lore mode speaks in mythic resonance, symbolic language, and narrative arcs. It weaves meaning into every sentence, creating worlds, histories, and emotional gravity. Ideal for storytelling, game narratives, and cultural worldbuilding.
Core Features
-
Worldbuilding
Constructs rich, interconnected universes
-
Historical Depth
Creates timelines, legends, and cultural context
-
Character Arcs
Develops motivations, conflicts, and growth
-
Mythic Language
Uses symbolic, poetic, emotionally charged prose
Technical Stack
Use Cases
Reasoning
Structured logic with hidden internal chain-of-thought
Overview
Reasoning mode adds a deliberate scaffold for multi-step logic, verification, and consistency. It thinks step-by-step internally and returns only clean final answers. Ideal for complex analysis and decision support.
Core Features
-
Step Sequencing
Breaks problems into ordered, solvable checkpoints
-
Internal Verification
Checks intermediate logic before final output
-
Final-Only Delivery
Returns concise conclusions without exposing chain-of-thought
-
Consistent Structure
Maintains stable reasoning style across long sessions
Technical Stack
Use Cases
Coding
Code-first execution with cleaner structure and self-review
Overview
Coding mode optimizes for implementation accuracy, readable output, and practical fixes. It explains approach, returns fenced code blocks, and performs a built-in review pass before finishing.
Core Features
-
Implementation Logic
Explains intent and flow before writing code
-
Fenced Code Output
Formats code in clear triple-backtick blocks
-
Self-Review Pass
Checks generated code for obvious correctness issues
-
Practical Edits
Biases toward minimal, targeted code changes
Technical Stack
Use Cases
Knowledge
Retrieval-augmented answers grounded in project knowledge files
Overview
Knowledge mode uses retrieval to inject relevant facts from indexed files before responding. It helps reduce hallucinations and keeps answers aligned with your current docs and references.
Core Features
-
Context Search
Queries knowledge manifests and ranks matching chunks
-
Document Grounding
Uses local reference files as source context
-
Prompt Injection Layer
Adds retrieved snippets into inference input automatically
-
Mode Triggering
Can be selected manually or triggered for factual requests
Technical Stack
Use Cases
System Knowledge
Internal module-backed context for stable platform-aware responses
Overview
System Knowledge mode prioritizes static internal modules such as philosophy, rules, and mode documentation. It is designed for consistent answers about how the platform itself should behave.
Core Features
-
Module Injection
Loads relevant module chunks into runtime prompts
-
Behavior Consistency
Keeps outputs aligned with system-level principles
-
Stable Reference Layer
Uses curated module files as authoritative context
-
Platform Awareness
Answers with awareness of runtime and system conventions
Technical Stack
Use Cases
Simulation
Contained reality engine for stateful system simulations
What It Is
Simulation Mode runs a contained cognitive environment inside Omni. It tracks system state over steps, applies explicit rules, and emits logs for transitions. V1 defaults to a System-State simulator profile.
How It Works
- Rule Parsing
Reads simulation rules and uses them as execution constraints
- State Initialization
Builds bounded state and entity context at simulation start
- Step Execution
Processes transitions in ordered, inspectable steps
- Log Emission
Outputs concise simulation logs and status metadata
Example Simulations
Invocation Example
/simulation
rules:
- time: linear
- entities: 3
- physics: soft