Archive History MCP server

How an agent would call it

Two steps: search returns refs, then data_fetch opens one of them. The arguments below are the arguments that were sent.

{
  "tool": "search",
  "arguments": {
    "query": "stripe.com/pricing",
    "sources": [
      "wayback"
    ],
    "limit": 2,
    "runId": "<from run_begin>"
  }
}

{
  "tool": "data_fetch",
  "arguments": {
    "sourceId": "wayback",
    "ref": {
      "id": "20260912141630:https://stripe.com/pricing"
    },
    "runId": "<from run_begin>"
  }
}

This runs against an account — send Authorization: Bearer <token> with the call. Get one at app.bedrocktap.com/dashboard/tokens; the free tier needs no card, and it reaches wayback like every other source.

What it costs. 0.25 credits per snapshot returned, and 0.25 per fetch. The search above returned 2 snapshots and cost 0.5 credits; the fetch cost 0.25.

What came back

{
  "ref": {
    "id": "20260912141630:https://stripe.com/pricing",
    "url": "https://web.archive.org/web/20260912141630/https://stripe.com/pricing",
    "title": "https://stripe.com/pricing @ 2026-09-12T14:16:30Z"
  },
  "text": "Snapshot of https://stripe.com/pricing captured 2026-09-12T14:16:30Z. Archived at https://web.archive.org/web/20260912141630/https://stripe.com/pricing",
  "fetchedAt": "2026-09-12T14:56:39.941Z"
}

Each ref carried timestamp, capturedAt, status, mimeType and original in its meta — both of the newest captures answered 200 as text/html. The fetched payload carried timestamp, capturedAt, original and archivedUrl.

Observed 2026-09-12. One call against a live source — ask again tomorrow and the rows move.

What a row contains

One row is one capture: an id of the form `<14-digit timestamp>:<url>`, the archived address, and a title reading `<url> @ <instant>`. meta carries the raw timestamp, the capture time as an ISO instant, the HTTP status the crawler recorded and the MIME type it saw. Rows come back newest first. A fetch returns the snapshot’s metadata and its archived address — never the archived page’s body, which belongs to whoever published it.

A question it answers. When did this competitor’s pricing page first appear, and how often has it changed since?

Routing guidance

What an agent reads to decide whether wayback is the right source, and what to reach for when it is not.

Answers

What a page said on a given date. Pricing before a change, a claim before it was quietly removed, a positioning statement before a pivot. The evidence layer under any "they used to say" argument.

Query shape

A URL and a date or date range. The nearest capture is returned when the exact date has none.

Price

Priced in credits, per row, per query or per call depending on the source. Every call is quoted before it runs — `cost_estimate` returns the exact figure and spends nothing. See /pricing.

Legal basis

Public API — the platform’s own documented, openly available interface.

Latency

Moderate.

Prefer instead

`page-diff` when the question is what CHANGED rather than what a page said at one moment.

Cannot answer

Pages never captured, pages excluded by their owner, and anything behind a login. Capture density is wildly uneven — a popular home page may have thousands of snapshots and a pricing page none.

Caveat

A snapshot may render incompletely because the archive did not capture its assets. Read the text, distrust the layout.

Related sources