dvorakmortonxd/freezmo
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Bring-your-own keys for cloud providers (OpenAI-compatible)

Browse source 
Settings -> Providers takes keys for OpenAI, Anthropic (their official
OpenAI-compatibility layer), DeepSeek, Groq, OpenRouter and any custom
OpenAI-compatible endpoint. One adapter (lib/openai-compat) covers
listing, streaming chat, JSON generation and vision, and every model
flow can run on a provider: the chat picker lists each enabled
provider's models (labelled, remembered), and a "Study generation"
choice moves decks, homework, grading and attachment transcription onto
a provider model.

Keys stay in this browser exactly like the Ollama one
(freezmo.providers.v1): never stored server-side, never in a backup or
the sync payload, forwarded only with the request that needs them. The
provider test asserts the key reaches the provider while the routing
header stays on the app side.

New suite tests/providers-test.mjs (13 checks) stubs both protocols:
model listing, chat streaming, study generation and the Settings flow.
Privacy policy updated (en+cs). Contrast 0 findings, axe clean, mobile
126/126, chat 102/102, wizard 95/95, bubbles 34/34, regression 44/44,
i18n 13/13, native 33/33. Version 0.4.7, SW -> freezmo-v72.
This commit is contained in:
Martin Dvořák 2026-10-14 06:29:39 +00:00
commit 4372423167
28 changed files with 1538 additions and 150 deletions
Download patch file Download diff file Expand all files Collapse all files

35
app/api/grade/route.ts
View file

@ -1,7 +1,8 @@
import { z } from "zod";
import { gradeShortAnswer } from "@/lib/ai";
import { authErrorMessage, authErrorKind, getServerConfig, resolveAuth } from "@/lib/config";
import { getServerConfig } from "@/lib/config";
import { INERT_OLLAMA_AUTH, resolveModelTarget } from "@/lib/providers-server";
import { friendlyErrorMessage, OllamaError } from "@/lib/ollama";
import { rejectCrossOrigin, rejectOversizedBody, requestLocale } from "@/lib/http-guard";
import { clientKey, rateLimit, rateLimitResponse } from "@/lib/rate-limit";
@ -12,6 +13,8 @@ export const runtime = "nodejs";
export const dynamic = "force-dynamic";
const requestSchema = z.object({
model: z.string().min(1).max(200).optional(),
question: z.string().min(1).max(1000),
referenceAnswer: z.string().min(1).max(2000),
keyPoints: z.array(z.string().max(300)).max(8).default([]),
@ -35,13 +38,14 @@ export async function POST(request: Request) {
const limit = rateLimit(`grade:${clientKey(request)}`, config.rateLimits.grade, config.rateLimits.windowMs);
if (!limit.allowed) return rateLimitResponse(limit, locale);
const auth = resolveAuth(config, request);
if ("error" in auth) {
return Response.json(
{ error: authErrorMessage(locale, auth.error), kind: authErrorKind(auth.error) },
{ status: 401 },
);
// Ollama (server-configured) or a cloud provider the browser sent a key for.
const resolved = resolveModelTarget(config, request, locale);
if ("error" in resolved) {
return Response.json({ error: resolved.error.message, kind: resolved.error.kind }, { status: resolved.error.status });
}
const target = resolved.target;
const provider = target.kind === "provider" ? target.target : undefined;
const baseAuth = target.kind === "provider" ? INERT_OLLAMA_AUTH : target.auth;
let payload: unknown;
try {
@ -55,12 +59,23 @@ export async function POST(request: Request) {
return Response.json({ error: translate(locale, "server.invalidGrade") }, { status: 400 });
}
// Provider runs carry the model the user picked in Settings → Providers.
const providerModel = provider ? parsed.data.model : undefined;
if (provider && !providerModel) {
return Response.json(
{ error: translate(locale, "server.providerModelNeeded"), kind: "bad_provider" },
{ status: 400 },
);
}
const textModel = provider ? providerModel! : config.textModel;
const gate = acquireModelGate(request, config, locale);
if ("response" in gate) return gate.response;
try {
const result = await gradeShortAnswer({
model: config.textModel,
auth: auth.auth,
model: textModel,
auth: baseAuth,
provider,
...parsed.data,
// Grading is quick; don't let a hung request block the feed for minutes.
signal: AbortSignal.timeout(Math.min(config.timeoutMs, 120_000)),
@ -70,7 +85,7 @@ export async function POST(request: Request) {
console.error("[api/grade]", error instanceof OllamaError ? `${error.kind}: ${error.message}` : error);
return Response.json(
{
error: friendlyErrorMessage(error, auth.auth.baseUrl, locale),
error: friendlyErrorMessage(error, baseAuth.baseUrl || provider?.baseUrl || '', locale),
kind: error instanceof OllamaError ? error.kind : "unknown",
},
{ status: 502 },