Tools & simulators

• Agent Loop Lab — Step through a real agent run one turn at a time — watch it think, call tools, observe, and decide when to stop.
• Is It an Agent? — Judge 18 real systems against the definition of an agent — chatbot, workflow, or the real thing, element by element.
• Architecture Advisor — Describe your problem, walk the decision tree, and get an architecture verdict — single agent, workflow, or multi-agent, with the tradeoffs stated.
• Orchestration Sandbox — Wire supervisor and worker agents together and watch tasks, tokens, and failures flow through the system.
• Protocol Picker — Function calling, MCP, or A2A? Answer questions about your integration and get a cited recommendation.
• Cloud Deployment Wizard — State your requirements and get the AWS, Azure, and GCP service map side by side — downloadable as CSV or Markdown.
• Well-Architected Lens Review — Review your agent against all 150 practices of the AWS Well-Architected Agentic AI Lens — filter by pillar and risk, track coverage, download the result.
• Prompt Injection Range — Attack a simulated agent yourself — direct injection, poisoned documents, tool-output smuggling — then flip sides and defend it.
• Tool Permission Lab — Scope an agent’s tool permissions, then watch what an attacker can still reach. Least privilege, made visceral.
• Trace Debugger — Real agent traces with one failing step hidden inside. Read the spans, find the failure, name the failure mode.
• Eval Suite Builder — Assemble an evaluation suite for a given agent — graders, datasets, thresholds — and see what your suite misses.
• Agent Incident Tabletop — An agent incident lands at 2 a.m. Triage it, contain it, communicate it — your choices decide the blast radius.
• AgentOps Maturity Assessment — Score your organization across the AgentOps dimensions and get a prioritized path to the next level.