Integrate LOLM into any platform

LOLM is an HTTP + SSE agent API with an npm client and a CLI. Use the hosted base URL or point at your own self-hosted instance — same routes, same receipts.

Quickstart (3 minutes)

npm install lolm-nfet-client

import { runCode, buildVisual, getStatus } from "lolm-nfet-client";

// optional: process.env.LOLM_BASE_URL || default public host
const baseUrl = "https://lolm.imagineqira.com";

console.log(await getStatus({ baseUrl }));

const done = await runCode({
  baseUrl,
  task: "write fizzbuzz to 20 in solution.py and run it",
  onEvent: (ev) => console.log(ev.event),
});
console.log(done?.receipt_sha || done?.receipt?.receipt_sha);

const vis = await buildVisual({ baseUrl, task: "a tiny counter app" });
// vis.html → render in iframe sandbox

npm install -g lolm-cli

lolm status
lolm code "write fizzbuzz to 20 in solution.py and run it" --save ./out
lolm build "a snake game" -o snake.html

# your own instance
lolm code "…" --base http://127.0.0.1:7866

# health
curl -sS https://lolm.imagineqira.com/api/demo/code/health | jq .

# usage remaining (no auth on free)
curl -sS https://lolm.imagineqira.com/api/demo/billing/usage | jq .

# coding agent (SSE stream)
curl -sSN -X POST https://lolm.imagineqira.com/api/demo/code/run \
  -H 'Content-Type: application/json' \
  -d '{"task":"print hello from main.py and run it"}'

# paid tier (after Stripe claim)
curl -sS https://lolm.imagineqira.com/api/demo/billing/usage \
  -H "X-LOLM-License: $LOLM_LICENSE"

import json, urllib.request

BASE = "https://lolm.imagineqira.com"

def get(path):
    with urllib.request.urlopen(BASE + path) as r:
        return json.loads(r.read().decode())

print(get("/api/demo/integrate")["endpoints"][:3])
print(get("/api/demo/billing/usage")["runs"])

# Streaming code run: use requests + iter_lines, or httpx,
# and parse SSE blocks (event: / data:). See JS parseSSEStream for the shape.

What you can call (public)

Nginx only exposes /api/demo/* on the public host. That is the stable integration surface.

Integration patterns

Auth & money (for your product)

• Free: no key — IP daily caps (8 runs / 2 visuals).
• Plus $9 / Pro $19: Stripe checkout → license token → send X-LOLM-License on every request.
• Memory scoping: optional X-Workspace-Owner: your-user-id.
• 402: daily quota exhausted — error body includes tiers, topups, and buy_topup. Offer plan upgrade (POST /api/demo/billing/checkout) or a one-time pack (POST /api/demo/billing/topup/checkout). Credits land via Stripe webhook + claim.

SSE shape (coding)

event: agent_note
data: {"text":"task state z_t opened…"}

event: command_started
data: {"cmd":"python3 solution.py"}

event: code_done
data: {"ok":true,"files":["solution.py"],…}

event: code_receipt
data: {"receipt_sha":"…","ok":true,"task":"…"}

Self-host base URL

# same client code — only base changes
const baseUrl = process.env.LOLM_BASE_URL || "http://127.0.0.1:7866";
await runCode({ baseUrl, task: "…" });

See install → self-host. CORS is open on the workspace app for browser demos; production backends should call from your server.

API keys (product identity)

Stable identity for integrators (better than IP free tier). Store the secret once — it is not shown again. This is not provider BYOK — those live on self-host under the Keys panel / /api/demo/keys.

curl -sS -X POST https://lolm.imagineqira.com/api/demo/api-keys \
  -H 'Content-Type: application/json' \
  -d '{"tier":"free","label":"my-app"}'

# use it
curl -sS https://lolm.imagineqira.com/api/demo/billing/usage \
  -H "X-LOLM-Api-Key: lolm_…"

# paid tiers: send X-LOLM-License from Stripe claim, then mint tier=plus|pro

BYOK (your model providers)

Self-host or loopback: paste Anthropic / OpenAI / Groq / Gemini / DeepSeek / Mistral / xAI / Fireworks / OpenRouter / Together / any OpenAI-compatible base URL. Keys apply instantly to chat, code, and visual builds. Your keys are preferred over the shared gateway.

# on your box (localhost)
curl -sS -X POST http://127.0.0.1:7866/api/demo/keys \
  -H 'Content-Type: application/json' \
  -d '{"keys":{"OPENAI_API_KEY":"sk-…","OPENAI_MODEL":"gpt-4o-mini"}}'

# prove it works
curl -sS -X POST http://127.0.0.1:7866/api/demo/keys/probe \
  -H 'Content-Type: application/json' \
  -d '{"provider":"openai"}'

Webhooks

Optional webhook_url on code runs — public HTTPS only (SSRF-guarded). Fired when the run completes with receipt summary.

curl -sSN -X POST https://lolm.imagineqira.com/api/demo/code/run \
  -H 'Content-Type: application/json' \
  -H "X-LOLM-Api-Key: $LOLM_API_KEY" \
  -d '{"task":"print 1 and run it","webhook_url":"https://your.app/hooks/lolm"}'

Python (stdlib)

from lolm_client import LOLM
c = LOLM(api_key="lolm_…")   # or LOLM() free IP
print(c.usage())
r = c.run_code_collect("write fizzbuzz to 20 in solution.py and run it")
print(r["done"], r["receipt"])

Module path: clients/python/lolm_client.py in the repo (PYTHONPATH=clients/python).