Context Engineering Tools
Your Context Engineering Tools
Interactive playgrounds for experimenting with context, agents, and models. Each one uses your inference budget — pick a model, see the cost, learn by doing.
Context Audit Workbench 9 stations
Token budget simulator, retrieval tuning, memory tier playground, injection tester with real AI, sampling playground, embedding explorer, constitutional AI testing, system patterns
Agent Loop Bench 5 stations + 8 demos
Configure OODA / ReAct / Plan-Execute, wire tools, set budgets, run a live agent loop step-by-step, 12-factor-agents interactive demos with token savings comparisons
Benchmark Bench 5 stations
Design test cases, write critic prompts, run simulated or real AI evaluations, compare models side-by-side, ICL comparison (zero-shot vs few-shot)
Omni Model Benchmarks live data
Live benchmark scores, pricing, and model recommendations for every major LLM. Compare models by task — coding, reasoning, search, agentic — with data from OpenRouter + arxiv.
Prompt Craft
Pattern library, prompt composition, A/B testing. Saves to your Portfolio.
Prompt Vocabulary
Directive wheel, aesthetic registers, prompt patterns. Saves to your Portfolio.
Context Strategy Planner
Session handoff templates, chunking strategies, context plans. Saves to your Portfolio.
Work you save to your Portfolio in any tool flows into the others — prompts built in Prompt Craft appear in the Context Workbench's token simulator.