Cognitive Layer

The Cognitive layer is the intelligence center — it handles natural language understanding, memory, learning, and self-modification.

Pipeline

Natural language goes through:

1. Working memory assembles system state (agent health, trust scores, Hebbian weights, capabilities) within a token budget
2. Episodic recall finds similar past interactions for context (top-3 by keyword-overlap cosine similarity)
3. Workflow cache checks for previously successful DAG patterns (exact match, then fuzzy)
4. LLM decomposer converts text into a TaskDAG — a directed acyclic graph of typed intents with dependencies
5. Attention manager scores tasks: urgency × relevance × deadline_factor × dependency_bonus
6. DAG executor runs independent intents in parallel, respects dependency ordering

Dynamic Intent Discovery

Each agent class declares structured IntentDescriptor metadata. The decomposer's system prompt is assembled at runtime from whatever agents are registered. New agent types self-integrate without any configuration changes.

This means adding a new agent type makes its intents available to the LLM automatically — no prompt editing, no routing tables, no configuration files.

Self-Modification

When ProbOS encounters a capability gap (no agent can handle a request), it designs a new agent:

Capability gap detected
    → LLM generates agent code
    → CodeValidator static analysis
    → SandboxRunner isolation test
    → Probationary trust assigned
    → SystemQA smoke tests
    → BehavioralMonitor tracks post-deployment

Agents can also be designed collaboratively via the /design command.

Correction Feedback Loop

Human corrections are the richest learning signal:

1. CorrectionDetector identifies when the user is correcting a previous result
2. AgentPatcher modifies the responsible agent
3. Hot-reload the patched agent
4. Auto-retry the original request
5. Update trust, Hebbian weights, and episodic memory

Dreaming

During idle periods, the dreaming engine:

• Replays recent episodes to strengthen successful pathways
• Weakens failed pathways
• Prunes dead connections
• Adjusts trust scores
• Pre-warms predictions for likely upcoming requests

Source Files

File	Purpose
cognitive/decomposer.py	NL → TaskDAG + DAG executor
cognitive/prompt_builder.py	Dynamic system prompt assembly
cognitive/llm_client.py	OpenAI-compatible + mock client
cognitive/cognitive_agent.py	Instructions-first LLM agent base
cognitive/working_memory.py	Bounded context assembly
cognitive/episodic.py	ChromaDB semantic long-term memory
cognitive/attention.py	Priority scoring + focus tracking
cognitive/dreaming.py	Offline consolidation + pre-warm
cognitive/workflow_cache.py	LRU pattern cache
cognitive/agent_designer.py	LLM designs new agents
cognitive/self_mod.py	Self-modification pipeline orchestrator
cognitive/code_validator.py	Static analysis for generated code
cognitive/sandbox.py	Isolated execution for untrusted agents
cognitive/skill_designer.py	Skill template generation
cognitive/skill_validator.py	Skill safety validation
cognitive/behavioral_monitor.py	Runtime behavior tracking
cognitive/feedback.py	Human feedback → trust/Hebbian/episodic
cognitive/correction_detector.py	Distinguishes corrections from new requests
cognitive/agent_patcher.py	Hot-patches designed agent code
cognitive/strategy.py	StrategyRecommender (skill attachment)
cognitive/dependency_resolver.py	Auto-install agent dependencies (uv)
cognitive/emergent_detector.py	5 algorithms for emergent behavior
cognitive/embeddings.py	Embedding utilities
cognitive/research.py	Web research phase for agent design