What problem does it solve?
Teams running LLM-based coding agents often pay Sonnet/Opus prices for simple code transforms that could run as zero-cost WASM transforms. This Skill reveals how much of your task batch qualifies for the $0 Tier 1 Agent Booster path so you can measure and reduce unnecessary LLM spend.
Core Features & Use Cases
- Batch task routing: Sends up to 50 task descriptions through the 3-tier
hooks_route router and captures each classification response.
- Tier partitioning: Splits tasks into Tier 1 (Agent Booster, $0 WASM transforms) versus Tier 2/3 (LLM) based on the
[AGENT_BOOSTER_AVAILABLE] marker, and extracts model recommendations.
- Cost reporting and persistence: Computes upper-bound spend for LLM-routed tasks, prints a bypass utilization report, and stores results in the
cost-tracking memory namespace for cost-report.
- Use Case: When
cost-report shows Sonnet spend on tasks like var-to-const or add-types, run this Skill on recent task descriptions to confirm how many could have bypassed the LLM entirely.
Quick Start
Run the cost-booster-route skill with --from-recent to analyze my recent task descriptions and show how many qualify for zero-cost Agent Booster routing.
