Agent Architecture
Design or review reliable AI agent systems. Use this whenever the user is creating an AI agent workflow, multi-agent system, agentic automation, orchestrator, tool-using assistant, subagent setup, or is having reliability problems with agents. Helps split responsibilities, limit context, separate planning from execution, gate side effects, define structured outputs, add validation, logs, handoffs…
System Bug Investigator
Investigate bugs, alerts, failed jobs, regressions, dirty worktree blockers, CI failures, deployment mismatches, and confusing system behavior before recommending or applying a fix. Use this when a problem may involve multiple files, scripts, services, schedulers, queues, databases, deployments, external tools, or runtime state. Default to diagnosis and the simplest safe first fix unless the user…