API & Integrations

Search API, MCP server, and plugin ecosystem for agents, CLI tools, and LLM integrations. Tokenless access with a small allowance, or sign up for a free API key with higher limits.

Authentication

An API key raises your rate limit from 2 requests/hour to 80 requests/30 minutes. Create your free key at /api/dashboard.

# Header auth (recommended)
curl "https://arlong.org/api/arlong/search?q=python" \
    -H "Authorization: Bearer al_xxxx"

# Query param auth
curl "https://arlong.org/api/arlong/search?q=python&key=al_xxxx"

import requests

r = requests.get(
    "https://arlong.org/api/arlong/search",
    params={"q": "python programming"},
    headers={"Authorization": "Bearer al_xxxx"}
)
data = r.json()

const res = await fetch(
    "https://arlong.org/api/arlong/search?q=javascript",
    { headers: { "Authorization": "Bearer al_xxxx" } }
);
const data = await res.json();

Search Live

Search the web and return structured results with relevance scores, domain reputation, and corroboration metadata.

Parameters

Response

curl "https://arlong.org/api/arlong/search?q=quantum+computing&pretty=1"

r = requests.get("https://arlong.org/api/arlong/search",
    params={"q": "quantum computing"})
results = r.json()["results"]
for r in results[:3]:
    print(f"{r['title']} — {r['url']}")

const { results } = await fetch(
    "https://arlong.org/api/arlong/search?q=quantum+computing"
).then(r => r.json());
results.slice(0, 3).forEach(r =>
    console.log(r.title, r.url)
);

Result fields

Answer Live

Ask a question and get a grounded AI answer with cited sources. The response includes the answer text, source list, and epistemic state.

Try it

Response

{
  "answer": "Retrieval-augmented generation (RAG) is a technique that...",
  "sources": [
    {"title": "...", "url": "https://...", "snippet": "..."}
  ],
  "epistemic_state": "4 independent domains examined; 3 contain a closely overlapping claim (not a factuality verdict)"
}

Security states

security_analysis.action is allow, review, block, or unknown. Review content is sanitised and may be used cautiously; blocked content is withheld. Unknown means no raw document was successfully scanned and can never be treated as allow.

Status Live

Live health snapshot of the model router and neural module.

Try it

Error Responses

MCP Server Live

Arlong exposes a full Model Context Protocol server, giving Claude Desktop, Cursor, and any MCP-compatible client access to web search, grounded AI answers, and live system status.

Remote Setup (Claude Desktop / Cursor)

Add to your MCP client config:

{
  "mcpServers": {
    "arlong": {
      "url": "https://arlong.org/mcp",
      "headers": {
        "Authorization": "Bearer al_xxxx"
      }
    }
  }
}

{
  "mcpServers": {
    "arlong": {
      "command": "npx",
      "args": ["mcp-remote", "https://arlong.org/mcp"],
      "env": {
        "ARLONG_API_KEY": "al_xxxx"
      }
    }
  }
}

Local Setup (stdio transport)

Run the MCP server as a local stdio process:

ARLONG_BASE_URL=https://arlong.org ARLONG_API_KEY=al_xxxx python mcp_arlong.py

Then point your MCP client at the script path:

{
  "mcpServers": {
    "arlong": {
      "command": "python",
      "args": ["/path/to/mcp_arlong.py"]
    }
  }
}

Available Tools

Tool schemas

{
  "name": "arlong_search",
  "description": "Search the web...",
  "inputSchema": {
    "type": "object",
    "properties": {
      "query": { "type": "string", "description": "The search query" },
      "page":  { "type": "integer", "default": 1 },
      "mode":  { "enum": ["instant", "balanced", "deep"] },
      "max_results": { "type": "integer", "maximum": 20 }
    },
    "required": ["query"]
  }
}

{
  "name": "arlong_extract",
  "inputSchema": {
    "type": "object",
    "properties": {
      "url": { "type": "string" },
      "query": { "type": "string" },
      "max_chars": { "type": "integer", "maximum": 12000 }
    },
    "required": ["url"]
  }
}

{
  "name": "arlong_answer",
  "description": "Ask a question and get a grounded AI answer...",
  "inputSchema": {
    "type": "object",
    "properties": {
      "query": { "type": "string", "description": "The question to answer" }
    },
    "required": ["query"]
  }
}

{
  "name": "arlong_status",
  "description": "Live health snapshot of the model router...",
  "inputSchema": { "type": "object", "properties": {} }
}

Quickstart

cURL

curl "https://arlong.org/api/arlong/search?q=weather+forecast&pretty=1"

Python

import requests

r = requests.get(
    "https://arlong.org/api/arlong/search",
    params={"q": "python programming"},
    headers={"Authorization": "Bearer al_xxxx"}
)
for res in r.json()["results"][:3]:
    print(f"{res['title']} — {res['url']}")

JavaScript

const res = await fetch(
    "https://arlong.org/api/arlong/search?q=javascript+fetch"
);
const { results } = await res.json();
console.log(results);

LLM function tool definition

{
  "type": "function",
  "function": {
    "name": "web_search",
    "description": "Search the web via Arlong. Add API key for higher limits.",
    "parameters": {
      "type": "object",
      "properties": {
        "q": { "type": "string", "description": "Search query" }
      },
      "required": ["q"]
    }
  }
}

Integrations & Plugins

Arlong integrates with the tools you already use.

Coming Soon

Best Practices

• Cache results — identical queries return the same results for one hour. Cache locally to avoid hitting the rate limit.
• Respect rate limits — 2 req/hour tokenless, or 80 req/30 minutes with an API key. Back off on 429 responses.
• Use the MCP server — for Claude Desktop / Cursor integrations, the remote MCP endpoint is the easiest path. Zero dependencies.
• Send your key — requests count against your account allowance instead of an anonymous IP limit. Free includes 100 API/MCP calls monthly; Founder and Pro include 2,000 per billing period. An 80-request/30-minute burst limit still applies.
• Monitor status — call /api/arlong/status or check the MCP arlong_status tool for router health before high-volume requests.