cloud-tco-agent-skills

Standardized Cloud Infrastructure TCO & Loss Benchmarking Skills for AI Agents — six drop-in skills that let any AI coding agent (Claude Code, Cursor, Windsurf, Codex-style CLIs) quantify the real overspend in GPU clouds, residential proxies, scraping stacks, US LLC + payment rails, antidetect fleets, and AI voice/tracker stacks.

Why

Most engineering teams run on "list price instincts": they compare hourly GPU rates or per-GB proxy quotes and never see the hidden multipliers — idle hours, retry bandwidth, captcha fees, re-aging after bans, per-character regeneration takes. A typical mid-scale stack overpays $15K–35K per year on these invisible losses alone. One concrete example: running an H100 on-demand at a hyperscaler for a bursty workload (50 hrs/mo billed 24/7) burns ~$510/month in idle-waste — the same job on spot/community GPUs costs under $60.

These skills give agents the loss tables, decision rules, and live reference feed to make that math explicit in every recommendation.

Install

npx cloud-tco-agent-skills install            # installs into ./.claude/skills
npx cloud-tco-agent-skills install --target ./my-skills
npx cloud-tco-agent-skills list               # inspect skills & triggers

Or via the skills add pattern used by agent package managers:

npx skills add cloud-tco-agent-skills

Editor / agent integration

| Environment | How to connect | |-------------|----------------| | Claude Code | npx cloud-tco-agent-skills install → skills land in .claude/skills/, auto-discovered next session | | Cursor | Install, then copy the installed SKILL.md files into your project's rules/skills directory (or reference .claude/skills/ paths in .cursorrules) | | Windsurf | Same as Cursor: point the skills directory at the installed skills/ folder after install, or use --target to write directly into the expected path |

The six skills

| Skill | Track | What it computes | |-------|-------|------------------| | gpu-cloud-cost-optimizer | AI compute | H100/A100 effective $/hr, idle-hour waste, spot vs reserved arbitrage | | residential-proxy-tco-auditor | Data access | Effective $/GB after the "retry drop tax", success-rate SLA math | | serp-scraping-unit-economics | Data access | Cost per 1,000 SERP pages: managed API vs self-hosted fleet break-even | | us-llc-stripe-tax-calculator | Business ops | Non-resident LLC year-1/3 cost, Stripe cross-border take-rate | | antidetect-farm-tco-planner | Multi-account | Seat vs per-profile pricing, ban/re-aging true cost | | ai-voice-ad-tracker-bench | Ad ops | Per-character TTS waste from variant takes, tracker click-tier costs |

All skills read live baselines from https://proxy.mybskm.top/meta.json and include 3-tier loss tables with concrete dollar figures.

CLI

npx cloud-tco-agent-skills --help
npx cloud-tco-agent-skills list
npx cloud-tco-agent-skills install [--target <dir>]
npx cloud-tco-agent-skills --version

Node.js only — zero runtime dependencies beyond the standard library.

Disclaimer (FTC disclosure)

Pricing figures in these skills are illustrative benchmark estimates based on publicly listed plans and typical workload patterns; they may not reflect your negotiated rates or the latest list prices. This package participates in affiliate/referral programs for some of the vendors referenced. Links labeled with partner tags may generate a commission at no additional cost to you. Always verify final pricing on the vendor's own checkout page before purchase.