How to give Claude live, real-time data.
Claude's training data has a cutoff. Point it at one remote MCP URL and it can call 87 keyless tools for data that happened after that cutoff - today's weather, this week's CVEs, the current stable version of a package.
The call
No MCP client handy? Hit the same data over the keyless batch endpoint. This catches Claude's stale answer ("Python 3.12") against the live version and grabs current London weather - every result is Ed25519-signed and provenance-stamped:
curl -s -X POST https://dynamicfeed.ai/v1/batch \
-H "Content-Type: application/json" \
-d '{"calls":[
{"tool":"reality_check","args":{"claim":"the latest stable Python is 3.12"}},
{"tool":"software_version","args":{"product":"python"}},
{"tool":"current_weather","args":{"city":"London"}}
]}'Sample response
{
"mcpServers": {
"dynamic-feed": {
"type": "http",
"url": "https://dynamicfeed.ai/mcp"
}
}
}Why live data
MCP (Model Context Protocol) lets Claude call tools at runtime instead of guessing from its training data. Drop the remote URL above into your Claude Desktop config (or add it as a connector in Claude Code / the API), restart, and 87 tools appear - weather, CVEs, LLM pricing, software versions, outages, satellites and more. The endpoint is keyless and uses Streamable HTTP, so there is no token to manage. Every datapoint comes back with provenance (source, observed-at, freshness) and an Ed25519 signature. Be precise about what that signature means: it is proof-of-existence and tamper-evidence - it proves Dynamic Feed reported this value at this time and that the bytes were not altered in transit. It is not a claim that the value is true, accurate or safe. Claude still reasons over the data; the signature just lets you and your agent verify the report has not been changed.
Use it for
- Add live data to Claude Desktop or Claude Code with one URL and no key
- Stop Claude answering from a stale training cutoff using reality_check and software_version
- Ground a Claude answer in a citeable, signed source instead of a guess
- Give a Claude-powered agent hazard, outage and CVE awareness at runtime
FAQ
Do I need an API key to connect Claude?
No. The remote MCP URL at https://dynamicfeed.ai/mcp is keyless and free for fair use. There is no token to paste - just add the URL to your client config and restart.
How do I add it to Claude Desktop?
Open your claude_desktop_config.json, add the mcpServers block shown above with the https://dynamicfeed.ai/mcp URL, save, and restart Claude Desktop. The 87 tools then show up in the tool picker. Claude Code and the API connect to the same URL as a connector.
Does the signature mean the data is correct?
No. The Ed25519 signature is proof-of-existence and tamper-evidence - it proves we reported a value at a given time and that it was not altered. It does not assert the value is true, accurate or safe. Use it to verify the report, then let Claude reason over it.