Migration assessment
Should I move? Per-target cost, exit egress, payback months, ranked recommendation. Triggers caveats automatically — e.g., "OCI A1.Flex is ARM, verify your AMIs".
assess_migration
Multi-cloud cost decisions across AWS, Azure, GCP, and OCI — migration assessment, RI break-even, 3-year TCO, egress arbitrage. Same engine that powers cloudprice-mcp on PyPI.
Every tool returns a structured artifact a FinOps team can sign off on — not estimates. Numbers traceable to a public catalog with explicit "what we don't model" disclosure on every result.
Should I move? Per-target cost, exit egress, payback months, ranked recommendation. Triggers caveats automatically — e.g., "OCI A1.Flex is ARM, verify your AMIs".
assess_migration
Six commitment scenarios from on-demand to 3yr-all-upfront with per-scenario monthly cost, upfront, and exact payback months.
optimize_commitment
Project per-cloud per-year cost with linear YoY growth assumptions. Returns cumulative TCO, year-by-year breakdown, and sensitivity analysis on the dominant variable.
compare_total_cost_of_ownership
Where do I save on data transfer? Surfaces the OCI 12× moat at 50 TB/month — $340 vs $4,000+ on hyperscalers. Includes payback math against one-time exit cost.
find_egress_arbitrage
Paste in a workload, get back per-cloud cost, exit egress, and payback months. Same engine, same numbers, no AI assistant required.
Every cloudprice catalog refresh writes a dated snapshot to the public repo. After enough weeks, you can answer questions no calculator can — "what did m5.xlarge cost in April?", "which OCI SKUs moved this month?" AWS / Azure / GCP / OCI calculators all show today only. We don't.
The v0.7 auto-refresh caught a 3.7× over-pricing in the previous hand-curated OCI catalog — first weekly snapshot already produced a real, useful finding. Each future refresh adds another data point.
Auto-generated from the public price-history dataset every deploy.
Endpoint: /history/biggest-movers.json ·
Per-SKU: /history/{cloud}/{sku}.json
Dataset is MIT-licensed and lives at
src/cloudprice_mcp/data/prices/ —
one JSON per weekly refresh, never overwritten. Also accessible from any MCP-compatible AI via the new
get_price_history and list_tracked_skus tools.
Backed by Claude 3 Haiku + the four cloudprice tools as function calls. Type a question, get back a structured FinOps answer with payback math and an audit trail of which tool was called.
Rate-limited 20 req/hr per IP. Every cost number traces to a public catalog. Tool-call audit shown below each reply.
Prefer a structured form? Use the migration / commitment / TCO / egress calculator below ↓
Auto-configures Claude Desktop, GitHub Copilot, Cursor, Windsurf, Cline, Continue.dev, and Zed. No manual config, no separate setup per client.
$ pip install cloudprice-mcp $ cloudprice-mcp setup # auto-detects every installed AI client; asks Y/N before writing each config $ cloudprice-mcp doctor ✓ Python version — 3.12.4 ✓ cloudprice-mcp package — v__CP_VERSION__ ✓ MCP tools registered — __CP_TOOL_COUNT__ tools