Agent Consciousness & Control Metalayer

Layers 1-2 of the Broomva Stack — 24 skills across 7 layers.

Skills for persistent consciousness in autonomous AI agent development. Three complementary skills that give stateless agent sessions the accumulated understanding of all prior sessions.

Skills

control-metalayer-loop — Behavioral Governance

Control-system metalayer with setpoints, sensors, gates, feedback loops, and escalation budgets.

npx skills add broomva/control-metalayer --skill control-metalayer-loop

agent-consciousness — Architecture & Philosophy

The synthesis: how control metalayer + knowledge graph + conversation logs form a self-evolving persistent consciousness.

npx skills add broomva/control-metalayer --skill agent-consciousness

knowledge-graph-memory — Episodic Memory Bridge

Bridges Claude Code conversation logs (.entire/ + transcripts) to an Obsidian knowledge graph for searchable session history.

npx skills add broomva/control-metalayer --skill knowledge-graph-memory

The Consciousness Stack

Working memory → Auto-memory → Conversation logs → Knowledge graph → Policy rules → Invariants
(ephemeral)                                                                        (permanent)

Substrate	Skill	Purpose
Control Metalayer	control-metalayer-loop	How to behave (gates, policies, setpoints)
Knowledge Graph	agent-consciousness	What is known (Obsidian vault, wikilinks, MOCs)
Conversation Logs	knowledge-graph-memory	What was done (session records, tool traces)

Quick Start

# 1. Initialize control metalayer
python3 .agents/skills/control-metalayer-loop/scripts/control_wizard.py init . --profile autonomous

# 2. Generate conversation history
python3 scripts/conversation-history.py --force

# 3. Audit
python3 .agents/skills/control-metalayer-loop/scripts/control_wizard.py audit . --strict

Self-Evolution

The system gets smarter the more it's used:

1. Agent encounters a failure mode not covered by existing policy
2. Agent fixes the immediate issue
3. Pattern is captured in conversation log (docs/conversations/)
4. If recurring, crystallizes into architecture doc (docs/architecture/)
5. If enforceable, becomes a gate in .control/policy.yaml
6. Future agents are governed by this rule automatically