Agent skill · data analytics · jeremylongshore
hex-cost-tuning
Optimize Hex costs through tier selection, sampling, and usage monitoring.
Why this skill is useful
Provides specific strategies and executable scripts for optimizing Hex API costs that the AI wouldn't reliably generate on its own.
What it needs
Requires hex account access. About 3k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Hex Cost Tuning Overview Hex pricing combines per-seat licensing with compute-based charges for notebook runs and data connections. Each scheduled or ad-hoc notebook execution consumes compute credits proportional to query complexity and data volume processed. Organizations running dozens of notebooks on hourly schedules — many producing identical results from unchanged data — accumulate unnecessary compute costs. Caching run results, optimizing schedules, and consolidating redundant notebooks are the highest-leverage cost reduction strategies. Cost Breakdown Component Cost Driver Optimization ----------- ------------ -------------- Seat licenses Per-user/month (Team: $28/user) Audit active editors quarterly; move viewers to free tier Notebook runs Compute per scheduled or manual execution Cache results for unchanged data; extend run intervals Data connections Active warehouse/database connections Consolidate overlapping connections; remove unused ones Scheduled runs Cron-triggered executions across all projects Audit schedules — reduce frequency for stable data API calls Admin and Run API requests Batch API operations; use cached results endpoint API Call Reduction Usage Monitoring Cost Optimization Checklist [ ] Cache notebook results with updateCacheResult: true [ ] Skip re-runs when source data is unchanged [ ] Audit scheduled run frequencies — extend intervals for stable data [ ] Move read-only users from Team to free viewer tier [ ] Consolidate duplicate notebooks querying the same data [ ] Remove unused data connections [ ] Set weekly run budget alerts at 80% threshold [ ] Identify overrun projects (>20 runs/week) for schedule review Error Handling Issue Cause Fix ------- ------- ----- Compute costs spiking Hourly schedules on notebooks with daily-changing data Extend schedule to match data refresh cadence Stale cached results Source data changed but cache not invalidated Use source data hash comparison before serving cache API rate limit (429) Too many concu …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/hex-cost-tuning--bcf176