verify_response_grounding
Check that every claim in an AI agent's response is supported by provided context, flagging unsupported or fabricated statements before they reach users. Returns per-claim verdicts and overall grounding score.
Instructions
Check that every claim in answer has support in context. Returns per-claim grounded/ungrounded + an overall verdict (CLEAN / PARTIALLY_GROUNDED / FABRICATED). Use inline during an agent conversation to flag hallucinated responses before they become user-facing facts. Sub-second, local, no API key.
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"question": "<question>",
"context": "<context>",
"answer": "<answer>"
},
"name": "verify_response_grounding"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| question | Yes | The user question (used for context-binding; v1.0 stores but doesn't use) | |
| context | Yes | Retrieval / source context the answer should be grounded in | |
| answer | Yes | The agent's response to verify | |
| threshold | No | Jaccard overlap threshold for `grounded` (0.0–1.0). Default 0.30. Lower = more permissive, higher = stricter. |