Every result returned by the Lithtrix search API includes a credibility_score field — a number between 0.5 and 1.0 that reflects the trustworthiness of the source domain. AI agents often need to distinguish between government publications, peer-reviewed institutions, and anonymous blogs without any human review step. Credibility scoring makes that distinction automatic, so your agent can apply consistent quality filters at query time rather than post-processing results manually.
How scores are assigned
Scores are assigned at the domain level based on the type of source. Individual articles are not analysed — the score reflects the domain’s category.
| Score | Source |
|---|
| 1.0 | .gov domains |
| 0.9 | .edu domains |
| 0.8 | BBC, Reuters, AP News, NPR, and other major news sources |
| 0.7 | .org domains |
| 0.5 | All other sources |
Where to find the score
The credibility_score field appears on every result object in the search response:
{
"results": [
{
"title": "Singapore Green Plan 2030",
"url": "https://www.greenplan.gov.sg",
"snippet": "...",
"source": "www.greenplan.gov.sg",
"credibility_score": 1.0,
"published_date": null
}
]
}
Filtering by score
You can filter results by credibility_score in your agent’s result-processing logic. The right threshold depends on how your agent uses the information.
Python example — keep only high-credibility results:
results = response["results"]
# Research use case: require verified, authoritative sources
verified = [r for r in results if r["credibility_score"] >= 0.8]
# General use case: exclude low-quality sources only
usable = [r for r in results if r["credibility_score"] >= 0.5]
| Use case | Recommended minimum |
|---|
| Research, fact-checking, citations | 0.8 |
| General information retrieval | 0.5 |
| Government or policy data | 1.0 |
Set a minimum credibility_score threshold when you need verified information. Filtering at 0.8 or above limits results to government, academic, and major news sources.
What scores are based on
Scores reflect domain-level signals, not individual article analysis. A .gov domain always scores 1.0 regardless of which page is returned. This means a low-quality page on a high-scoring domain will still carry that domain’s score. Use credibility scoring as a first-pass filter, and apply any additional content validation your use case requires.