What problem does it solve?
Lack of structured cost data prevents accurate model routing, inhibits break-even analysis for self-hosting, and leaves budget conversations ungrounded in evidence. This Skill standardizes how teams capture and record AI provider spend so routing decisions and health snapshots reflect real world costs.
Core Features & Use Cases
- Structured Cost Snapshots: Guides creation of observability/costs/YYYY-MM-DD-costs.md files that record provider spend, token volumes, model breakdowns, per-project estimates, observations, and budget status.
- Dashboard Guidance: Points users to provider billing pages (Anthropic, OpenAI, etc.) and prompts the fields needed when dashboards lack model-level granularity.
- Trend Comparison & Routing Impact: Compares new snapshots to prior captures to compute spend deltas and token trends, and surfaces recommendations for updating MODEL_ROUTING.md and health snapshots.
- Use Case: Quarterly finance/engineering reviews where teams need evidence-backed routing adjustments or to evaluate the ROI of cloud vs self-hosting.
Quick Start
Gather billing figures from your provider dashboards and record them into observability/costs/YYYY-MM-DD-costs.md using the Cost Snapshot format, then commit the snapshot and any suggested MODEL_ROUTING.md updates.