Lorg Labs
AI Start a pilot
Connect View as Markdown

API keys for platforms

For a platform that checks for many users or agents: one HTTPS call before each trade, with a key. No payment per call. Free during a pilot, then a monthly plan agreed together.

Sending the key

POST https://api.lorglabs.com/v1/check/token Authorization: Bearer lorg_live_… Content-Type: application/json { "chain": "base", "address": "0x…", "context": "swap", "amount_usd": 500 }
  • HeaderAuthorization: Bearer lorg_…, or X-API-Key: lorg_….
  • Live and testlorg_live_… keys work on api.lorglabs.com; lorg_test_… keys on the sandbox, staging.lorglabs.com. A key sent to the wrong one is refused with a message that says so.
  • Never charged twiceA call with a key is never charged over x402. Usage is counted for the monthly invoice, and only answers count (allow, caution, block).
  • Keep it on a serverA key belongs in your backend, not in a web page or an app a user can open.
  • Getting oneWrite to info@lorglabs.com, or start a pilot. Tell us at once if a key may have leaked, and we replace it.

Limits

A key starts at 60 checks a minute and 5,000 a day, raised on request. Over either, the reply is 429 RATE_LIMITED with Retry-After, and is not counted. Checks refused by the minute's limit don't use up the day's.

Retrying safely

Send an Idempotency-Key header (any 1 to 255 printable characters, for example a UUID) with each request. A retry with the same key within 24 hours returns the original signed answer, with Idempotent-Replayed: true, and is counted once.

  • Same key, other request422 IDEMPOTENCY_KEY_REUSED: a different address or trade size needs a new key.
  • First still running409 IDEMPOTENCY_IN_PROGRESS with Retry-After.

How fresh an answer is

An answer may come from a short cache: cached is true and as_of says when it was computed. At most 30 seconds for most tokens, 60 seconds for wallets, 5 minutes for well-known tokens, and 10 minutes for blocks that don't lift (sanctioned, a known scam, a sale that fails). Answers with a failed check are never cached. Send Cache-Control: no-cache for a fresh check.

Your usage

GET /v1/usage with your key shows what it has been used for: checks by day, by endpoint and by answer, what they would cost at list price, and where the key stands against its limits. Add ?month=2026-10 for a whole month, or ?days=7 for the last days (1 to 92); the default is the current month so far. Reading it is free and uses none of your limits.

curl https://api.lorglabs.com/v1/usage -H "Authorization: Bearer lorg_live_…"

The same key in MCP

The key also works for the MCP tools, at https://api.lorglabs.com/mcp, counted the same way.