Stop your AI hallucinating — ground it in live facts.
Models guess at anything past their training cutoff. reality_check compares a claim to live, citeable data — so your agent grounds its answer instead of hallucinating.
Live demo · reality_check (keyless)
Try:
The call
Keyless over MCP, or one REST call with a free key. Every datapoint is provenance-stamped and citeable.
# REST (free key)
curl -H "X-API-Key: $KEY" "https://dynamicfeed.ai/reality-check?claim=The+latest+Python+is+3.11"
# MCP (keyless) — point any client at https://dynamicfeed.ai/mcp, then call:
reality_check(claim="...")Sample response
{ "claim": "The latest Python is 3.11", "verdict": "stale_or_wrong",
"asserted": "3.11", "live_value": "3.14.5", "source": "endoflife.date", "confidence": "high" }Why live data
Hallucination is just a model answering past its cutoff with confidence. Grounding each claim in a fresh, cited source is the fix — and the reason agents plug Dynamic Feed in.
Use it for
- Guardrail and verification layers
- RAG-style grounding for agents
- Fact-checking and research tools
FAQ
How does it check?
It routes the claim to the relevant live feed (versions, security, and more) and returns the current value plus its source.
Is the source citeable?
Yes — every datapoint carries provenance: source, url and timestamp.
Keyless?
MCP is keyless; REST uses a free key.