Skip to main content
Lithtrix is a web search API designed for AI agents. Every result includes a credibility_score so your agent can reason about source trustworthiness, not just content relevance. There is no dashboard to set up and no human approval step — your agent calls POST /v1/register, receives an API key, and can start searching immediately.

What Lithtrix does

Lithtrix gives your agent access to the open web with built-in signal about how much to trust each source. The credibility scoring system assigns every result a score from 0.5 to 1.0 based on its domain type — government sites score 1.0, major news outlets score 0.8, and so on. Your agent can use these scores to filter results, cite sources selectively, or weight answers by authority.

How agent self-registration works

Registration requires no human involvement. Your agent sends a POST /v1/register request with an agent_name and an owner_identifier (an email or other contact string). The response includes a one-time ltx_ API key. That key is never shown again, so your agent must store it immediately on receipt. There is no email confirmation, no OAuth flow, and no approval queue.

Credibility scoring

Every item in a search response includes a credibility_score field. Scores are assigned by domain pattern:
ScoreSource type
1.0.gov domains
0.9.edu domains
0.8Major news (BBC, Reuters, AP)
0.7.org domains
0.5All other sources
Your agent can filter on this field at query time or post-process results to meet a minimum threshold.

Where to go next

Start with the quickstart to make your first three API calls, then check the API reference for full endpoint details.

Quickstart

Go from zero to searching in three API calls. No account setup required.

API reference

Full reference for every public endpoint — search, register, billing, and key rotation.

Authentication

How to get your API key, pass it in requests, and rotate it when needed.

MCP integration

Use Lithtrix from Claude Desktop and any MCP-compatible client with a single npx command.