What problem does it solve? Understanding how AI agents consume tokens, hit caches, and trigger anomalies across a shared better-ccflare proxy is difficult without a structured, read-only reporting workflow that respects privacy and data limits. ## Core Features & Use Cases - Usage and cost analysis: Queries the remote better-ccflare Insights API over Tailscale to report per-agent request counts, token breakdowns, cost, latency, and failover metrics for a chosen time range. - Cache and anomaly reporting: Surfaces global cache hit rates, savings when pricing is known, and aggregate anomaly detector findings with redacted identifiers. - Privacy-safe advisory output: Produces a fixed-section advisory report with coverage caveats, never printing prompts, request IDs, or raw JSON, and never modifying proxy configuration. - Use Case: Ask for a 7-day ccflare usage review to see which agents drive token spend, whether cache reuse is effective, and whether any runaway loops need attention. ## Quick Start Ask the assistant to run a ccflare-insights report for the last 7 days summarizing agent usage, token cost, and cache behavior.
