Pricing Time Machine

What the Pricing Time Machine MCP server does

Pricing Time Machine MCP server retrieves archived copies of SaaS pricing pages from the Wayback Machine and turns them into structured comparisons. It can show how prices changed, identify periods with identical pricing, report plans or prices that appeared or disappeared, and return the page text surrounding extracted amounts.

The server fits the Search & Data Extraction category. It is useful for researching competitor pricing, reconstructing historical plan changes, and comparing several companies at a shared point in time. It does not scrape live websites and does not require API keys.

How it works

A tool request first checks the Wayback CDX index to find archived copies, then selects the copy closest to the requested date. Dates may be supplied as a year, a year and month, a full date, or latest; the selected archive date is included in the result.

The server performs snapshot lookup, price extraction, and grouping of repeated prices. Extracted information can include amount, currency, monthly or annual billing, per-user wording, a guessed plan name, nearby context, and page text around prices. Interpretation remains with the MCP client or model, including deciding whether a price belongs to a particular plan or what a change means.

Archived responses are cached on disk at ~/.cache/pricing-time-machine. The cache can reduce repeated archive requests, while the first request may take 10–30 seconds per page because archive.org can be slow. Retries with backoff are used when the archive is temporarily unavailable.

Setup and configuration

Pricing Time Machine MCP server requires uv. For Claude Code, add it with:

claude mcp add pricing-time-machine -- uvx pricing-time-machine-mcp

Claude Desktop and Cursor can launch it through an MCP configuration using uvx as the command and pricing-time-machine-mcp as its argument. The README does not specify required environment variables or API credentials.

For local development, the repository documents uv sync --extra dev followed by uv run pytest. Its tests use a mocked Wayback Machine, while a separate smoke script checks the live archive.

Tools and capabilities

• pricing_timeline samples archived copies across years, groups unchanged prices into periods, and lists additions and removals.
• pricing_at finds prices near a requested date and includes plan guesses, billing hints, context, and surrounding page text.
• compare_pricing compares one pricing page at two dates with both price tables and detected additions or removals.
• compare_competitors compares two to eight pricing pages at the same date.
• pricing_snapshots reports which months or years have archived copies.

Two documented prompts support broader workflows: pricing_history_report requests a dated history for one company, and competitor_pricing compares a market across time. Pricing Time Machine MCP server can therefore support questions such as identifying historical price increases, checking an old plan price, or comparing several SaaS products in a given year.

Limitations and notes

Results depend on Wayback Machine coverage. A page may have no useful archive for the requested period, even though many popular pricing pages have monthly captures. Pages that render prices only with JavaScript may appear empty in the archive; trying a nearby date may help.

Plan names are inferred from nearby headings rather than guaranteed metadata. The returned context and page text should be checked before treating a plan assignment as certain. Archive availability and response time can also vary because archive.org may be temporarily offline or slow.