Find where an agent pipeline can swap frontier models for cheaper ones without losing quality. Ingests the user's agent trace logs, replays successful steps through cheaper models on OpenRouter, and uses an LLM-as-judge to check the cheaper output matches the original. Handles multi-step / tool-calling / looping pipelines by re-executing the real code with the model swapped, then reports per-step cost savings, quality scores, and cascade risks in a TUI + report. Use when the user wants to cut model spend, right-size models per task, benchmark cheaper model substitutions, or analyze an agentic pipeline (LangGraph, Codex, Claude Code, raw SDK) for cost optimization. Not a prompt-improvement or prompt-rewriting tool — it changes which model runs a step, never the prompt.
2026-07-16
