feat: initial cog release — cognitive architecture for Claude Code

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

memory/cog-meta/improvements.md

@@ -0,0 +1,8 @@
+# Cog — Improvements
+<!-- L0: Feature ideas, wishlists, and repair notes -->
+
+<!-- Edit in place by section. -->
+
+## Ideas
+
+## Implemented

memory/cog-meta/patterns.md

@@ -0,0 +1,4 @@
+# Cog — Learned Patterns
+<!-- L0: Distilled interaction and workflow rules from experience -->
+
+<!-- Edit in place. Distill self-observations into actionable patterns. Timeless rules only. HARD LIMIT: 110 lines / 7.5KB. -->

memory/cog-meta/scenario-calibration.md

@@ -0,0 +1,14 @@
+# Scenario Calibration
+<!-- L0: Prediction accuracy tracker for decision scenarios -->
+
+<!-- Updated by /reflect when scenarios resolve. -->
+
+## Resolved Scenarios
+
+| Scenario | Created | Resolved | Predicted Branch | Actual Branch | Accuracy | Lesson |
+|----------|---------|----------|-----------------|---------------|----------|--------|
+
+## Metrics
+
+- Total resolved: 0
+- Accuracy rate: —

memory/cog-meta/self-observations.md

@@ -0,0 +1,4 @@
+# Cog — Self-Observations
+<!-- L0: What worked and didn't in interactions — append-only -->
+
+<!-- Append-only. Format: - YYYY-MM-DD [tag]: observation -->