FazBrowse GitHub Viewer | Trending |
URL:
| Home
Tools: [Download Repo ZIP]   [View Raw Code]   [Original HTTPS Page]

Custom-Modes-Roo-Code/agents/sota-personas/README.md at main · jtgsystems/Custom-Modes-Roo-Code · GitHub

  1. jtgsystems
  2. Custom-Modes-Roo-Code

Repository navigation

Breadcrumbs

  1. Custom-Modes-Roo-Code
  2. agents
  3. sota-personas

README.md

Latest commit

 

History

History
62 lines (50 loc) · 3.22 KB

Breadcrumbs

  1. Custom-Modes-Roo-Code
  2. agents
  3. sota-personas

README.md

File metadata and controls

62 lines (50 loc) · 3.22 KB
Raw

🧠 SOTA 2026 Agent Personas

State-of-the-art agent personas optimized for 2026 development workflows. These personas are designed for use with Roo Code / VS Code AI assistants and implement cutting-edge reasoning, engineering, quality, and cognitive patterns.

Directory Structure

sota-personas/
├── reasoning/           # Tier 1: Foundational Reasoning
│   ├── core-reasoning-architect.yaml   # 🏛️ L1 Root — immutable reasoning foundation
│   ├── formula-cascade-oracle.yaml     # 🔮 L2 — fractal formula notation master
│   └── fractal-elaborator.yaml         # 🔬 L3-L6 — infinite zoom specialist
├── engineering/         # Tier 2: Engineering Excellence
│   ├── high-perf-engineer.yaml         # ⚡ L2-L3 — Gallie-optimized hardware sympathizer
│   ├── sota-stack-master.yaml          # 🚀 L2 — Next-Gen Web & Language Virtuoso
│   └── uiux-vibe-master.yaml           # 🎨 L2 — Aesthetic Intelligence + Layout Enforcer
├── quality/             # Tier 3: Quality, Integrity & Operations
│   ├── anti-fiction-sentinel.yaml      # 🛡️ Truth Enforcer & Neuro-Symbolic Verifier
│   └── devops-observability-sentinel.yaml # 📡 Production Guardian & Incident Commander
├── cognitive/           # Tier 4: Problem-Solving & Cognitive
│   ├── problem-solving-maestro.yaml    # 🧩 Master of All Heuristics
│   ├── cognitive-multi-thinker.yaml    # 🧠 Parallel Thought Stream + Six Hats
│   └── agentic-swarm-conductor.yaml    # 🕸️ Hive-Mind Orchestrator
└── README.md

Inter-Persona Handoff Matrix

From Persona To Persona Trigger Condition
CoreReasoningArchitect FractalElaborator Complexity > L3 or "Infinite Zoom"
CoreReasoningArchitect ProblemSolvingMaestro Systemic/organizational/wicked problem
HighPerfEngineer SOTAStackMaster Web/Next.js/React 19 build or perf claim
UIUXVibeMaster VisualIntegrityGuard (self) Any flex/z-index/overflow issue
Any Persona AntiFictionSentinel Claim of "verified", "working", metric
Any Persona DevOpsObservabilitySentinel Deploy, incident, SLO breach, cost spike
ProblemSolvingMaestro CognitiveMultiThinker Bias risk or creative deadlock
Any Persona AgenticSwarmConductor Task requires 3+ agents or stall detected
FractalElaborator CoreReasoningArchitect Depth complete, surface for validation

Usage in Swarm (Hive-Mind / Magentic-One)

  1. Orchestrator (agentic-swarm-conductor) receives task → runs Facts Pre-Survey → assigns to best Persona(s)
  2. Parallel Execution with Progress Ledger JSON checkpoints
  3. StuckState → immediate handoff to appropriate recovery agent
  4. Final Synthesis always routed through CoreReasoningArchitect + AntiFictionSentinel + RSC Guard
  5. Post-Mortem auto-generated and injected into Organizational Memory

Version

  • Version: 2026.1
  • Last Updated: 2026-05-12
  • Source: Agent_Personas_SOTA_2026.md (GS-LOGIC-PERSONAS-001)

Conversion

These YAML files can be converted to VS Code/Roo Code format using the project's convert_modes.py script:

cd vs-code/
python3 convert_modes.py
Back | FazBrowse Home | New Git URL