@nerua1

nerua1-ultrawork runs multiple independent tasks in parallel by routing each to a tiered model based on complexity, collecting results when all complete. targets 3x speedup on batch work without sequential blocking.

nerua1-arena-council executes multiple local llms in parallel and aggregates responses via majority or weighted voting. targets cost reduction and robustness over single-model inference.

Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope,...