broomva Control Metalayer
Three complementary skills that give stateless AI agent sessions the accumulated understanding of all prior sessions.
Stack position: Layer 1 (Foundation) + Layer 2 (Memory & Consciousness) of the Broomva Stack — 24 skills across 7 layers.
Included Skills
| Skill | Layer | Purpose |
|---|---|---|
control-metalayer-loop |
Foundation (L1) | Control primitives: setpoints, sensors, gates, feedback loops, escalation budgets, and behavioral governance |
agent-consciousness |
Memory (L2) | Architecture and philosophy for persistent agent consciousness across sessions |
knowledge-graph-memory |
Memory (L2) | Episodic memory bridge from Claude Code conversation logs to an Obsidian knowledge graph |
Install
# Install all skills
npx skills add broomva/control-metalayer
# Install individual skills
npx skills add broomva/control-metalayer --skill control-metalayer-loop
npx skills add broomva/control-metalayer --skill agent-consciousness
npx skills add broomva/control-metalayer --skill knowledge-graph-memory
How They Fit Together
Control Metalayer (how to behave)
↕
Agent Consciousness (what is known)
↕
Knowledge Graph Memory (what was done)
Information flows from ephemeral working memory through conversation logs into the knowledge graph, where recurring patterns crystallize into permanent control policies. Each new agent session inherits the full governance and memory substrate from all prior sessions.
Stack Integration
Higher layers depend on this package:
| Layer | Skills | How they use control-metalayer |
|---|---|---|
| Orchestration (L3) | symphony, autoany | Dispatch governed by policy gates |
| Research (L4) | deep-dive-research | Research traces persisted to knowledge graph |
| Strategy (L7) | drift-check, decision-log, weekly-review | Decisions and reviews write to vault via knowledge-graph-memory; drift-check reads control-metalayer setpoints |