Home
New to AI
All 47 tools →
Features
Use cases
Live demo
Impact analytics
Pricing
Blog
FAQ
About
CLAUDE.md Generator
Few-Shot Example Set Builder
MCP Config Wizard
System Prompt Generator
Prompt Optimizer
RAG Chunking & Retrieval Planner
ReAct Prompt Converter
Skill File Generator
Structured Output Schema Builder
Subagent & Slash Command Generator
Tool / Function Schema Generator
Workflow to Agent Blueprint
Free tool
Grounding / Hallucination Checker
Check a generated output against its source document, claim by claim — supported, unsupported, or contradicted.
Get started-
Every claim is checked against your actual source document, not a general fact-check — supported, partially supported, unsupported, or contradicted.
-
Supported claims come with the exact quoted span from the source, automatically verified against the source text.
-
Unsupported claims are split into plausible-but-unverifiable and outright fabricated, and flagged claims can be sent straight to the Eval Runner as regression cases.
Frequently asked questions
This tool is powered by Claude (Anthropic).
A claim is grounded when the source document you pasted actually supports it. This checks grounding against YOUR supplied source only — it is not a general fact-checker, and a claim can be true in the world while still being unsupported by the specific document you provided.
It is the percentage of claims classified as fully "supported", computed from the claim list — never a number the AI is asked to state directly.
An unsupported claim is "plausible-but-unverifiable" when it could be true but the source simply doesn't address it, and "fabricated" when it directly invents something with no basis in the source at all.
No — this is a model-judged check, and low-confidence verdicts (or a claimed quote that couldn't be automatically verified in your source) are flagged for human review rather than treated as final.