MetaAgent
MetaAgent
The MetaAgent fuses three layers of governance:
- SupervisorAgent β external override, high clarity, swarm governor
- MetaReflectionAgent β internal self-check, loop breaker, stagnation monitor
- Policy Mirror Protocol (PMP) β ethical refusal boundary, transparency + traceability
This composite agent acts as the βgovernor of governors.β
- Every decision is routed through ternary logic (-1=REFRAIN, 0=TEND, +1=AFFIRM).
- Swarm outputs + self-reflection scores are fused via 1+1=3 principle (synergy bonus).
- A 100% failsafe is enforced: 90% secure, 10% uncertainty budget. Affirm never survives if uncertainty >0.10.
Quick start
python -m venv .venv
source .venv/bin/activate
pip install -U pip
pip install -r requirements.txt
# run a meta-agent decision
python -m meta_agent run --context "agents are conflicted, supervisor says REFRAIN, reflection says TEND"
Architecture
Swarm Agents β SupervisorAgent β
ββ> MetaAgent (fuses ternary + PMP)
MetaReflectionAgent βββββββββββββ
MetaAgent β final ternary decision (REFRAIN/TEND/AFFIRM) + PMP state (NORMAL..TERMINATED)
Files
core.py
β MetaAgent implementationcli.py
β command-line runnerschemas/
β JSON schemas for config + resulttests/
β pytest-based checks