feat(agent): Define session-less cognitive architecture

- **Anatomy (Data Stores):** Defines the three-tiered data model:
  1. Linguistic Substrate (Org files)
  2. Lisp Object Store (Live graph in RAM)
  3. Telemetry Store (External TSDB for sensory data)

- **Psychology (Cognitive Matrix):** Establishes the 2x2 matrix that
  balances Conscious/Subconscious and Neural/Symbolic (System 1/2)
  processing.

- **Physiology (Core Processes):** Outlines the five core processes
  (Perception, Reasoning, Distillation, Reflection, Sensation) that
  bring the architecture to life.

This commit also updates the master GTD with the full engineering
roadmap required to implement this session-less vision, including tasks
for cognitive economics, belief revision, and the Micro-Prolog engine.

README.org

@@ -39,6 +39,29 @@ A Common Lisp microkernel that maintains a live, threaded Object-Store in RAM. I
 ** Gemini CLI
 A strategic orchestrator that handles large-scale batch tasks, research, and planning. It operates according to the *PSF Consensus Loop* (Demand -> Blueprint -> Success -> Build -> Chaos -> Memory).
 
+* The Cognitive Architecture: A Session-less Mind
+The agent's "mind" is not a transient chat session but a durable, stateful cognitive architecture. This design enables a truly "session-less" experience, where the agent's context, personality, and awareness are as persistent as the data itself.
+
+*** The Anatomy: Three Data Stores
+1. *The Linguistic Substrate (Org Files):* The human-readable Source of Truth, comprising the "Journal" ([[file:daily/README.org][Dailies]]) and the "Ledger" ([[file:notes/README.org][Notes]]).
+2. *The Lisp Object Store (RAM):* The "Active Brain," a live graph of Lisp objects representing every headline, paragraph, and entity in the Memex, with vectors embedded directly inside.
+3. *The Telemetry Store (TSDB/Binary):* An external, high-volume database for sub-symbolic "Sensory" data (e.g., smart home logs), which is monitored and distilled by the agent.
+
+*** The Psychology: The 2x2 Cognitive Matrix
+The agent's "thought" process is not monolithic. It operates on a 2x2 matrix that balances cognitive speed with cognitive state, allowing for sophisticated, multi-layered reasoning.
+
+| | *System 1 (Neural/Fast/Intuitive)* | *System 2 (Symbolic/Slow/Logical)* |
+| :--- | :--- | :--- |
+| *Conscious (Foveal/Active)* | *The Interface:* A fast, charming LLM for active conversation. | *The Pilot:* Lisp guards that ensure active requests are safe. |
+| *Subconscious (Peripheral/Passive)* | *The Editor:* A smart, deep LLM that finds patterns in the background. | *The Librarian:* A deterministic Lisp engine that files, archives, and maintains the integrity of memory. |
+
+*** The Physiology: Five Core Processes
+1. *Perception (The Attention Bridge):* Automatically vectorizes user input to set the "Foveal Focus" for the active conversation.
+2. *Reasoning (The Micro-Prolog):* Uses symbolic logic to reconcile contradictions and enforce the "Physics" of the Memex.
+3. *Distillation (The Scribe):* The Subconscious loop that turns the daily "Journal" entries into permanent "Ledger" knowledge.
+4. *Reflection (The Gardener):* The heartbeat-driven process that finds "forgotten" links and maintains the "Vibe" of the system.
+5. *Sensation (The Sensor Skill):* The "Signal-to-Symbol" converter that turns the "flood" of telemetry into actionable =TODO= items.
+
 * Core Workflows: The Life of a Thought
 1. *Capture:* Raw information enters via the `inbox` (tasks) or `daily` logs (thoughts).
 2. *Distill (The Scribe):* Automated agents periodically extract conceptual thoughts from the chronological logs into evergreen, atomic notes.