Integration Guide

AgenticCognition integrates with other Agentra sisters through typed bridge traits. All bridges have NoOp defaults, so no external dependency is ever required.

Bridge Architecture

Each bridge is a Rust trait with a NoOp default implementation. When a sister is available, the real implementation replaces the NoOp. This pattern ensures AgenticCognition compiles and runs without any sister present.

pub trait MemoryBridge: Send + Sync {
    fn recall_context(&self, topic: &str) -> Option<Vec<String>> { None }
    fn store_insight(&self, insight: &str) -> Result<()> { Ok(()) }
}

Memory Bridge

Connect AgenticMemory to provide historical conversation context. Memory observations strengthen belief evidence and improve shadow detection accuracy.

Capabilities when connected:

• Retrieve past conversation topics to inform belief confidence
• Store cognition insights as memory facts for cross-session recall
• Link belief changes to specific conversation episodes
• Use temporal chains to track when beliefs were formed or challenged

Planning Bridge

Connect AgenticPlanning to feed goals and decision commitments into the decision fingerprint analysis. Planning state helps predict future choices.

Capabilities when connected:

• Import active goals as context for decision simulation
• Feed commitment status into drift tracking
• Use goal completion patterns to refine the prediction engine
• Detect value-goal misalignment

Identity Bridge

Connect AgenticIdentity for signed model snapshots and tamper-evident verification of .acog files.

Capabilities when connected:

• Sign model snapshots with cryptographic receipts
• Verify model integrity through the identity chain
• Attribute belief changes to specific interaction sessions
• Enable trust-gated access to sensitive shadow data

Vision Bridge

Connect AgenticVision to incorporate visual observations into the user model.

Capabilities when connected:

• Use UI interaction patterns as behavioral evidence
• Detect visual attention patterns that reveal preferences
• Incorporate screenshot-based observations into belief evidence

Codebase Bridge

Connect AgenticCodebase to model the user's technical identity and coding patterns.

Capabilities when connected:

• Import coding style patterns into the decision fingerprint
• Use code review preferences to strengthen work-domain beliefs
• Track technical growth through codebase evolution

Comm Bridge

Connect AgenticComm to model communication patterns and social dynamics.

Capabilities when connected:

• Analyze message tone and style as behavioral evidence
• Track communication preference drift over time
• Detect projection patterns in inter-agent communication

Cognition Bridge (self-bridge)

AgenticCognition exposes its own bridge trait for other sisters to consume.

Capabilities exposed:

• Predict user preferences for any sister's decision points
• Provide belief context for memory storage decisions
• Offer shadow awareness for identity trust calculations

Standalone Usage

AgenticCognition is independently installable and operable. No sister integration is required for any core functionality. The full feature set -- belief physics, shadow detection, drift tracking, prediction -- works entirely standalone.

# Install standalone
curl -fsSL https://agentralabs.tech/install/cognition | bash

# Use without any sister
acog model create
acog belief add $MODEL_ID "I prefer clarity over cleverness" --domain values
acog shadow map $MODEL_ID

MCP Integration

The agentic-cognition-mcp server exposes 14 tools over the Model Context Protocol. Any MCP-compatible client gains instant access to persistent user modeling. See MCP Tools for the complete tool reference.

Standalone Guarantee

AgenticCognition is independently installable and operable. Sister integrations enhance capability but are never required for core functionality. The NoOp bridge pattern ensures that missing sisters never cause errors -- they simply return neutral defaults.