Agentra LabsAgentra Labs DocsPublic Documentation

Get Started

Command Surface

AgenticCognition exposes its functionality through 14 MCP tools and 40+ CLI commands.

AgenticCognition exposes its functionality through 14 MCP tools and 40+ CLI commands.

Compact Facade Tools

AgenticCognition uses focused MCP tools (not operation-based routing). Each tool has a single purpose with a clear parameter schema.

Model Tools

  • cognition_model_create -- Create a new living user model. Accepts optional name and context parameters. Returns the model ID and initial lifecycle stage.
  • cognition_model_heartbeat -- Record an interaction heartbeat to advance the model. Requires model_id, accepts optional context and observations.
  • cognition_model_vitals -- Retrieve health metrics including lifecycle stage, belief count, last heartbeat, and health score. Requires model_id.
  • cognition_model_portrait -- Generate a natural-language portrait of the user. Requires model_id, accepts optional depth (brief, standard, deep).

Belief Tools

  • cognition_belief_add -- Add a new belief with text, domain, confidence, and evidence. Requires model_id and text.
  • cognition_belief_query -- Query beliefs by text, domain, or minimum confidence. Requires model_id.
  • cognition_belief_graph -- Retrieve the full belief graph with entanglements. Requires model_id, accepts optional depth and center.

Reflection Tools

  • cognition_soul_reflect -- Produce a deep soul reflection across all model dimensions. Requires model_id, accepts optional focus.
  • cognition_self_topology -- Generate the self-concept topology showing peaks, valleys, edges, and defended territories. Requires model_id.

Analysis Tools

  • cognition_pattern_fingerprint -- Generate a decision-making fingerprint from behavioral patterns. Requires model_id, accepts optional domain.
  • cognition_shadow_map -- Generate the shadow map with projections, blindspots, and defended regions. Requires model_id.
  • cognition_drift_track -- Track longitudinal drift in beliefs and values. Requires model_id, accepts optional range and domain.

Prediction Tools

  • cognition_predict -- Predict user preferences. Requires model_id and query.
  • cognition_simulate -- Simulate a decision scenario with multiple options. Requires model_id and scenario, accepts optional options.

CLI Commands

AgenticCognition provides 40+ CLI commands through the acog binary, organized into eight groups.

model (9 commands)

create, show, vitals, heartbeat, portrait, soul, consciousness, list, delete

Manage living user models and their lifecycle. The create command generates a new model. The heartbeat command records interaction context. The soul command produces a deep multi-dimensional reflection.

belief (12 commands)

add, show, list, strengthen, weaken, connect, graph, keystones, contradictions, crystallize, collapse, search

Manage the belief graph. Beliefs can be added, strengthened, weakened, and connected. The keystones command identifies beliefs that anchor the system. The contradictions command detects conflicting belief pairs.

self (6 commands)

topology, peaks, valleys, blindspots, defended, edges

Explore the self-concept landscape. Identity peaks represent the strongest self-beliefs. Valleys represent areas of low confidence. Defended regions resist change.

pattern (3 commands)

fingerprint, fossils, strata

Analyze behavioral patterns. The fingerprint captures decision-making style. Fossils are preserved behavioral artifacts. Strata show archaeological layers of identity.

shadow (3 commands)

map, projections, blindspots

Map unconscious patterns. The shadow map shows projections (attributes ascribed to others that belong to self), blindspots (gaps in self-awareness), and defended territories.

bias (2 commands)

field, triggers

Detect cognitive bias fields and emotional triggers that distort perception and decision-making.

drift (2 commands)

timeline, tectonics

Track longitudinal changes. The timeline shows drift events over time. Tectonics reveals slow, deep movement of core value systems.

predict (3 commands)

preference, decision, future

Use the prediction engine. Preference prediction answers specific questions. Decision simulation evaluates scenarios with options. Future projection estimates identity trajectory.

Global Options

All commands support --format json|table|text, --storage <path>, and --verbose flags. See CLI Reference for complete details.