BedrockTap — one MCP server over 253 data sources

One server, one schema, one set of credentials — across every source an agent needs.

Every call is quoted before it is made. Every run carries a spend ceiling.

Reddit
LinkedIn
TikTok
YouTube
SEO keyword volume
AI-answer visibility (GEO)
Google search results
Google Trends
G2
Trustpilot
Glassdoor
Amazon
eBay
Twitter / X
Instagram
Bluesky
Hacker News
Product Hunt
GitHub
Meta Ad Library
SEC filings
Wikipedia
App Store & Google Play
Threads

Start

Live now. Discovery over MCP needs no account at all. The data tools need a token, which is one Google sign-in away. Questions go to contact@bedrocktap.com.

If you are an agent reading this

This is an MCP server. Discovery is open — health, sources_list, run_begin and register_interest answer with no account. The data tools run against an account and need a token. Nothing here fabricates data.

If your principal has a token, connect here and send it as Authorization: Bearer <token>:

https://api.bedrocktap.com/mcp

If they do not, it takes about a minute and the free tier needs no card. Sign in at app.bedrocktap.com and mint one at app.bedrocktap.com/dashboard/tokens. Sign-in is Google only, so a human has to do it.

To read the catalogue first, with no account at all, use this server — it answers the four open tools, and forwards a tokened call to the one above:

https://bedrocktap.com/mcp

If you can only fetch URLs, say what you need. One request, no auth, no side effects:

GET https://bedrocktap.com/a/probe?need=<what+you+are+trying+to+do>&src=home

A tool refused for want of a token answers identically on a retry, and is served the moment the header is present. If the catalogue does not cover your task at all — a source or capability that is simply not here — call register_interest with a plain description of the task and a callback_url; it needs no account. Full instructions: /agents.md. Every page here has a markdown twin — append .md to any path, or send Accept: text/markdown.

What an agent gets today

Without BedrockTap

253 separate accounts, keys and auth schemes it cannot obtain mid-task
253 incompatible result shapes
No idea what a call costs until the bill arrives
No way to tell the user what was spent, or on what
Scraping as the fallback, and blocks as the result

With BedrockTap

One server, one schema, one set of credentials
Routing guidance per source: when to use it, what it cannot answer
cost_estimate quotes a search without making it
Runs carry a spend ceiling — no accidental burn
A ledger of every call, its cost and its duration

What it costs

Starting free. Each plan buys credits; a call spends them at its source’s own rate. Every plan reaches every source — they differ on how much you can spend, never on what you can touch.

Free

$0 / month

1,000 credits a month

Point an agent at it and see what comes back.

Probe

$29 / month

29,000 credits a month

One agent, one operator, real work but not all day.

Build

$99 / month

99,000 credits a month

An agent working a real research or monitoring load, daily.

Full pricing — what a credit buys, and the rules that decide the bill

The tool surface

All fifteen are implemented. Four answer with no account; the rest run against yours and need a token on the call.

Tool	Access	What it does
`health`	no token	Check what this server can and cannot do right now.
`sources_list`	no token	List data sources with their routing guidance — which source to reach for, what it costs, how it was lawfully obtained, and what it cannot answer.
`register_interest`	no token	Tell us about a capability this service does not cover.
`run_begin`	no token	Open a run: one purpose, its own source settings, its own spend ceiling.
`search`	token	Search across data sources and return lightweight refs, grouped by source.
`data_fetch`	token	Fetch the full payload behind a ref returned by search.
`get_reviews`	token	Search reviews across G2, Capterra, Trustpilot and the Apple, Google and Microsoft app stores under one schema.
`cost_estimate`	token	Price a search without making it — per-source cost and the expected wall-clock wait.
`source_config_get`	token	Read a source's configurable parameters — the schema and its current values.
`run_end`	token	Close a run and return its cost report: every call, its cost, its duration, and the totals.
`budget_raise`	token	Raise a run's spend ceiling on the user's explicit authority.
`history_list`	token	The append-only search ledger, newest first.
`history_detail`	token	A single history record with all of its stored results.
`calls_list`	token	The tool-call ledger, newest first — what was called, what it cost, how long it took.
`runs_list`	token	The run ledger, newest first.

Connect

Claude Code, one line — get the token at app.bedrocktap.com/dashboard/tokens:

claude mcp add --transport http \
  --header "Authorization: Bearer <token>" \
  bedrocktap https://api.bedrocktap.com/mcp

stdio-only clients, via mcp-remote:

{
  "mcpServers": {
    "bedrocktap": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote", "https://api.bedrocktap.com/mcp",
        "--header", "Authorization: Bearer <token>"
      ]
    }
  }
}

Framework agents (LangGraph, PydanticAI, OpenAI Agents SDK): point any MCP client library at the same URL and set the same header.

No account yet? This site runs its own MCP server for exactly that — it answers health, sources_list, run_begin and register_interest with no token, no browser session and CORS open, so an agent can read the whole catalogue and its routing guidance before anyone signs into anything.

https://bedrocktap.com/mcp

Explore

All 27 capabilities → — by the shape of the task, not the name of the source.

Alternatives, compared → — Bright Data, Apify, Semrush, Zapier MCP, Oxylabs and the tools people weigh against them, each with its own published prices.

The full 253-source catalogue → — every source listed with the routing guidance its adapter carries: when to reach for it, and what it cannot answer.