birdclaw 0.7.0 → 0.8.0

package/CHANGELOG.md

@@ -1,5 +1,21 @@
 # CHANGELOG
 
+## 0.8.0 - 2026-06-10
+
+### Added
+
+- Add localized Today/digest reports through `--language <locale-id>`, `BIRDCLAW_DIGEST_LANGUAGE`, and the period-digest API, with canonical locale validation and separate caches. (#47 - thanks @yujiawei)
+
+### Changed
+
+- Update runtime and development dependencies, align TanStack Start packages, and move pnpm's native build allowlist into workspace configuration.
+
+### Fixed
+
+- Add accurate archive-first Sign in and archive-request onboarding, including account-binding requirements, current xurl/bird setup, scoped transport selection, and clean autolink rendering. (#46 - thanks @peetzweg)
+- Respect `OPENAI_BASE_URL` when sending Today/digest requests to OpenAI-compatible API endpoints.
+- Anchor Link Insights' Today range to UTC midnight so it matches stored `created_at` timestamps across local time zones.
+
 ## 0.7.0 - 2026-06-01
 
 ### Added

package/README.md

@@ -219,6 +219,8 @@ birdclaw auth status --json
 birdclaw db stats --json
 ```
 
+`auth status` reports Birdclaw's coarse xurl status. Verify xurl with `xurl whoami` and bird with `bird whoami`. For setup and transport selection, see [Sign in](docs/auth.md).
+
 Find and import an archive:
 
 ```bash
@@ -227,6 +229,8 @@ birdclaw import archive --json
 birdclaw import archive ~/Downloads/twitter-archive-2025.zip --json
 ```
 
+Don't have an archive yet? Request it from <https://x.com/settings/download_your_data>; X emails a download link when it is ready, which may take a few days. A fresh Birdclaw database needs the archive import to establish account identity before live sync. See [Archive Import → Get an archive](docs/archive.md#get-an-archive).
+
 Optional profile hydration can improve bios, follower counts, and avatars, but it performs live X profile reads and can spend API credits on large archives:
 
 ```bash
@@ -414,17 +418,18 @@ birdclaw profile-analyse openai --max-pages 20 --max-conversations 40 --conversa
 
 ### What happened today
 
-`birdclaw today` streams a local "what happened" digest from the SQLite store. It uses the OpenAI Responses API with `gpt-5.5`, medium reasoning, and priority service tier by default. Set `OPENAI_API_KEY`; override with `BIRDCLAW_AI_MODEL`, `BIRDCLAW_OPENAI_REASONING_EFFORT`, or `BIRDCLAW_OPENAI_SERVICE_TIER` when needed.
+`birdclaw today` streams a local "what happened" digest from the SQLite store. It uses the OpenAI Responses API with `gpt-5.5`, medium reasoning, and priority service tier by default. Set `OPENAI_API_KEY`; override with `BIRDCLAW_AI_MODEL`, `BIRDCLAW_OPENAI_REASONING_EFFORT`, or `BIRDCLAW_OPENAI_SERVICE_TIER` when needed. Use `--language <locale-id>` or `BIRDCLAW_DIGEST_LANGUAGE` for localized reports.
 
 ```bash
 birdclaw today
+birdclaw today --language zh-CN
 birdclaw digest 24h --refresh
 birdclaw digest week --json
 birdclaw digest --since 2026-05-16T00:00:00Z --until 2026-05-17T00:00:00Z
 birdclaw digest today --include-dms
 ```
 
-The web UI exposes the same stream under `What happened`. DMs are excluded unless explicitly enabled. Final structured results are cached by the exact local context hash, model, reasoning effort, and service tier.
+The web UI exposes the same stream under `What happened`. DMs are excluded unless explicitly enabled. Final structured results are cached by the exact local context hash, model, reasoning effort, service tier, and report language.
 
 ### Search and triage DMs
 

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "birdclaw",
-	"version": "0.7.0",
+	"version": "0.8.0",
 	"description": "Local Twitter memory in SQLite for archives, DMs, likes, bookmarks, and moderation",
 	"homepage": "https://github.com/steipete/birdclaw#readme",
 	"license": "MIT",
@@ -56,53 +56,47 @@
 		"@tailwindcss/vite": "^4.3.0",
 		"@tanstack/devtools-vite": "0.7.0",
 		"@tanstack/react-devtools": "0.10.5",
-		"@tanstack/react-router": "1.169.2",
-		"@tanstack/react-router-devtools": "1.166.13",
-		"@tanstack/react-router-ssr-query": "1.166.12",
-		"@tanstack/react-start": "1.167.65",
-		"@tanstack/router-plugin": "^1.167.35",
-		"@vitejs/plugin-react": "^6.0.1",
-		"commander": "^14.0.3",
-		"effect": "^3.21.2",
-		"kysely": "^0.29.0",
-		"lucide-react": "^1.16.0",
+		"@tanstack/react-router": "1.170.15",
+		"@tanstack/react-router-devtools": "1.167.0",
+		"@tanstack/react-router-ssr-query": "1.167.1",
+		"@tanstack/react-start": "1.168.25",
+		"@tanstack/router-plugin": "^1.168.18",
+		"@vitejs/plugin-react": "^6.0.2",
+		"commander": "^15.0.0",
+		"effect": "^3.21.3",
+		"kysely": "^0.29.2",
+		"lucide-react": "^1.17.0",
 		"mapbox-gl": "^3.24.0",
-		"react": "^19.2.6",
-		"react-dom": "^19.2.6",
+		"react": "^19.2.7",
+		"react-dom": "^19.2.7",
 		"react-map-gl": "^8.1.1",
 		"supercluster": "^8.0.1",
 		"tailwindcss": "^4.3.0",
-		"tsx": "^4.22.0",
-		"vite": "^8.0.13",
+		"tsx": "^4.22.4",
+		"vite": "^8.0.16",
 		"zod": "^4.4.3"
 	},
 	"devDependencies": {
 		"@playwright/test": "^1.60.0",
-		"@tailwindcss/typography": "^0.5.19",
+		"@tailwindcss/typography": "^0.5.20",
 		"@testing-library/dom": "^10.4.1",
 		"@testing-library/jest-dom": "^6.9.1",
 		"@testing-library/react": "^16.3.2",
 		"@types/geojson": "^7946.0.16",
-		"@types/node": "^25.8.0",
-		"@types/react": "^19.2.14",
+		"@types/node": "^25.9.2",
+		"@types/react": "^19.2.17",
 		"@types/react-dom": "^19.2.3",
 		"@types/supercluster": "^7.1.3",
-		"@typescript/native-preview": "7.0.0-dev.20260514.1",
-		"@vitest/coverage-v8": "^4.1.6",
+		"@typescript/native-preview": "7.0.0-dev.20260609.1",
+		"@vitest/coverage-v8": "^4.1.8",
 		"jsdom": "^29.1.1",
-		"oxfmt": "^0.49.0",
-		"oxlint": "^1.64.0",
+		"oxfmt": "^0.54.0",
+		"oxlint": "^1.69.0",
 		"typescript": "^6.0.3",
-		"vitest": "^4.1.6"
+		"vitest": "^4.1.8"
 	},
 	"engines": {
 		"node": ">=25.8.1 <27"
 	},
-	"packageManager": "pnpm@10.31.0",
-	"pnpm": {
-		"onlyBuiltDependencies": [
-			"esbuild",
-			"lightningcss"
-		]
-	}
+	"packageManager": "pnpm@10.31.0"
 }

package/scripts/build-docs-site.mjs

@@ -24,7 +24,10 @@ const productDescription =
 const brewInstall = "brew install steipete/tap/birdclaw";
 
 const sections = [
-	["Start", ["index.md", "install.md", "quickstart.md", "configuration.md"]],
+	[
+		"Start",
+		["index.md", "install.md", "auth.md", "quickstart.md", "configuration.md"],
+	],
 	[
 		"Archive & Sync",
 		["archive.md", "sync.md", "media.md", "backup.md", "jobs.md"],
@@ -449,19 +452,41 @@ function inline(text, currentRel) {
 		stash.push(`<code>${escapeHtml(code)}</code>`);
 		return `\u0000${stash.length - 1}\u0000`;
 	});
-	out = escapeHtml(out)
-		.replace(/\*\*([^*]+)\*\*/g, "<strong>$1</strong>")
-		.replace(/(^|[^*])\*([^*\s][^*]*?)\*(?!\*)/g, "$1<em>$2</em>")
-		.replace(/(^|[^_])_([^_\s][^_]*?)_(?!_)/g, "$1<em>$2</em>")
-		.replace(
-			/\[([^\]]+)\]\(([^)]+)\)/g,
-			(_, label, href) =>
-				`<a href="${escapeAttr(rewriteHref(href, currentRel))}">${label}</a>`,
-		)
-		.replace(/&lt;(https?:\/\/[^\s<>]+)&gt;/g, '<a href="$1">$1</a>');
+	const stashLink = (label, href) => {
+		stash.push(
+			`<a href="${escapeAttr(rewriteHref(href, currentRel))}">${renderInlineText(label, stash)}</a>`,
+		);
+		return `\u0000${stash.length - 1}\u0000`;
+	};
+	out = out.replace(/\[([^\]]+)\]\(<([^<>]+)>\)/g, (_, label, href) =>
+		stashLink(label, href),
+	);
+	out = out.replace(/\[([^\]]+)\]\(([^)]+)\)/g, (_, label, href) =>
+		stashLink(label, href),
+	);
+	out = out.replace(/<(https?:\/\/[^\s<>]+)>/g, (_, url) => {
+		stash.push(`<a href="${escapeAttr(url)}">${escapeHtml(url)}</a>`);
+		return `\u0000${stash.length - 1}\u0000`;
+	});
+	return renderInlineText(out, stash);
+}
+
+function renderInlineText(text, stash) {
+	let out = formatInlineText(text);
 	out = out.replace(/\\\|/g, "|");
 	out = out.replace(/&lt;br&gt;/g, "<br>");
-	return out.replace(/\u0000(\d+)\u0000/g, (_, i) => stash[Number(i)]);
+	return restoreInlineStash(out, stash);
+}
+
+function formatInlineText(text) {
+	return escapeHtml(text)
+		.replace(/\*\*([^*]+)\*\*/g, "<strong>$1</strong>")
+		.replace(/(^|[^*])\*([^*\s][^*]*?)\*(?!\*)/g, "$1<em>$2</em>")
+		.replace(/(^|[^_])_([^_\s][^_]*?)_(?!_)/g, "$1<em>$2</em>");
+}
+
+function restoreInlineStash(text, stash) {
+	return text.replace(/\u0000(\d+)\u0000/g, (_, i) => stash[Number(i)]);
 }
 
 function rewriteHref(href, currentRel) {
@@ -886,6 +911,8 @@ function highlightYamlValue(rest) {
 	return escapeHtml(rest);
 }
 
+export const __test__ = { inline };
+
 function validateLinks(outputDir) {
 	const failures = [];
 	const placeholderHrefs =

package/src/cli.ts

@@ -63,6 +63,7 @@ import {
 	syncMentions,
 } from "#/lib/mentions-live";
 import {
+	normalizeDigestLanguage,
 	streamPeriodDigest,
 	type PeriodDigestOptions,
 	type PeriodDigestPreset,
@@ -371,6 +372,7 @@ function buildDigestOptions(
 		account?: string;
 		includeDms?: boolean;
 		model?: string;
+		language?: string;
 		refresh?: boolean;
 		since?: string;
 		until?: string;
@@ -398,6 +400,14 @@ function buildDigestOptions(
 	if (liveSyncMode === undefined) {
 		return null;
 	}
+	let language: string | undefined;
+	try {
+		language = normalizeDigestLanguage(options.language);
+	} catch (error) {
+		printError(error instanceof Error ? error.message : String(error));
+		process.exitCode = 1;
+		return null;
+	}
 	return {
 		period: parseDigestPeriod(period),
 		since: options.since,
@@ -406,6 +416,7 @@ function buildDigestOptions(
 		includeDms: Boolean(options.includeDms),
 		refresh: Boolean(options.refresh),
 		model: options.model,
+		language,
 		maxTweets,
 		maxLinks,
 		liveSync: options.liveSync !== false,
@@ -1268,6 +1279,10 @@ program
 	.option("--account <accountId>", "Account id")
 	.option("--include-dms", "Include private DM context")
 	.option("--model <model>", "OpenAI model id")
+	.option(
+		"--language <tag>",
+		"Report language as a Unicode locale id, e.g. zh-CN (env: BIRDCLAW_DIGEST_LANGUAGE)",
+	)
 	.option("--refresh", "Bypass the local digest cache")
 	.option("--max-tweets <n>", "Maximum tweet context", "5000")
 	.option("--max-links <n>", "Maximum linked articles", "12")
@@ -1292,6 +1307,10 @@ program
 	.option("--since <isoDate>", "Start of explicit window")
 	.option("--until <isoDate>", "End of explicit window")
 	.option("--model <model>", "OpenAI model id")
+	.option(
+		"--language <tag>",
+		"Report language as a Unicode locale id, e.g. zh-CN (env: BIRDCLAW_DIGEST_LANGUAGE)",
+	)
 	.option("--refresh", "Bypass the local digest cache")
 	.option("--max-tweets <n>", "Maximum tweet context", "5000")
 	.option("--max-links <n>", "Maximum linked articles", "12")

package/src/lib/link-insights.ts

@@ -380,6 +380,8 @@ function resolveRange(
 
 	const start = new Date(now);
 	if (range === "today") {
+		// "Today" follows the user's local calendar day; toISOString below
+		// converts that local midnight into the correct UTC storage boundary.
 		start.setHours(0, 0, 0, 0);
 	} else {
 		const days = range === "week" ? 7 : range === "month" ? 30 : 365;

package/src/lib/period-digest.ts

@@ -28,6 +28,7 @@ export interface PeriodDigestOptions {
 	includeDms?: boolean;
 	refresh?: boolean;
 	model?: string;
+	language?: string;
 	reasoningEffort?: "minimal" | "low" | "medium" | "high";
 	serviceTier?: "default" | "flex" | "priority";
 	signal?: AbortSignal;
@@ -108,6 +109,32 @@ const PeriodDigestSchema = z.object({
 	sourceTweetIds: z.array(z.string()).default([]),
 });
 
+const MAX_DIGEST_LANGUAGE_LENGTH = 64;
+
+export function normalizeDigestLanguage(
+	value: string | undefined,
+): string | undefined {
+	const trimmed = value?.trim();
+	if (!trimmed) return undefined;
+	if (
+		trimmed.length > MAX_DIGEST_LANGUAGE_LENGTH ||
+		!/^[a-z0-9]+(?:-[a-z0-9]+)*$/i.test(trimmed)
+	) {
+		throw new Error(
+			"Digest language must be a valid Unicode locale identifier such as en, zh-CN, or pt-BR",
+		);
+	}
+	try {
+		const [canonical] = Intl.getCanonicalLocales(trimmed);
+		if (!canonical) throw new Error("missing canonical locale");
+		return canonical;
+	} catch {
+		throw new Error(
+			"Digest language must be a valid Unicode locale identifier such as en, zh-CN, or pt-BR",
+		);
+	}
+}
+
 export type PeriodDigest = z.infer<typeof PeriodDigestSchema>;
 
 interface CompactTweet {
@@ -563,6 +590,12 @@ export function collectPeriodDigestContext(
 	};
 }
 
+function languageFromOptions(options: PeriodDigestOptions) {
+	return normalizeDigestLanguage(
+		options.language ?? process.env.BIRDCLAW_DIGEST_LANGUAGE,
+	);
+}
+
 function modelFromOptions(options: PeriodDigestOptions) {
 	return options.model ?? process.env.BIRDCLAW_AI_MODEL ?? DEFAULT_MODEL;
 }
@@ -834,16 +867,23 @@ function digestCacheKey(
 	context: PeriodDigestContext,
 	options: PeriodDigestOptions,
 ) {
-	return [
+	const parts = [
 		"period-digest:v2",
 		modelFromOptions(options),
 		reasoningEffortFromOptions(options),
 		serviceTierFromOptions(options),
 		context.hash,
-	].join(":");
+	];
+	const lang = languageFromOptions(options);
+	if (lang) parts.push(`lang:${lang}`);
+	return parts.join(":");
 }
 
-function buildPrompt(context: PeriodDigestContext) {
+function buildPrompt(
+	context: PeriodDigestContext,
+	options?: { language?: string },
+) {
+	const language = normalizeDigestLanguage(options?.language);
 	const promptTweets = context.tweets.map((tweet) => ({
 		id: tweet.id,
 		url: tweet.url,
@@ -928,7 +968,7 @@ Requirements:
 - Stream one readable Markdown report first. The UI will show this text directly; do not rely on separate cards or structured summaries.
 - Target 700-1100 words when there is enough data.
 - Start with a 2-3 sentence lead that immediately says what people are talking about.
-- Use sections named "What people are talking about", "Important links shared", and "Worth opening". Add "Worth replying to" only if there are clearly high-signal replies.
+- Use sections named "What people are talking about", "Important links shared", and "Worth opening". Add "Worth replying to" only if there are clearly high-signal replies. Translate these section titles when a report language is requested.
 - When a tweet has replyToTweet, use that parent context to understand what the author was replying to and whether Peter already joined the conversation.
 - Use bullets under each section. Each bullet should be specific and explain why it matters.
 - For tweets: cite every claim with inline tweet ids at the end of the relevant sentence or bullet, e.g. (tweet_123, tweet_456). These citations become hoverable source links.
@@ -942,6 +982,7 @@ Requirements:
 - Keep actionItems empty unless you wrote a "Worth replying to" section.
 - Put every tweet id cited in the Markdown into sourceTweetIds.
 - JSON shape: { "title": string, "summary": string, "keyTopics": [{ "title": string, "summary": string, "tweetIds": string[], "handles": string[] }], "notableLinks": [{ "title": string, "url": string, "why": string, "sourceTweetIds": string[] }], "people": [{ "handle": string, "name"?: string, "why": string }], "actionItems": [{ "kind": "reply"|"follow_up"|"read"|"sync", "label": string, "tweetId"?: string, "dmConversationId"?: string }], "sourceTweetIds": string[] }
+${language ? `- Write all human-readable prose, including section titles and JSON prose fields, in ${language}.\n- Preserve handles, URLs, tweet ids, and JSON property names exactly.` : ""}
 
 Dataset:
 ${JSON.stringify(dataset)}`;
@@ -950,12 +991,22 @@ ${JSON.stringify(dataset)}`;
 function fallbackDigest(
 	context: PeriodDigestContext,
 	markdown: string,
+	language?: string,
 ): PeriodDigest {
-	const title = `${context.window.label} digest`;
+	const normalized = markdown.replaceAll(/\s+/g, " ").trim();
+	const heading = markdown
+		.split("\n")
+		.map((line) => line.match(/^#{1,6}\s+(.+)$/)?.[1]?.trim())
+		.find(Boolean);
+	const neutralFallback = language ? `[${language}]` : undefined;
 	return {
-		title,
+		title:
+			heading?.slice(0, 160) ??
+			neutralFallback ??
+			`${context.window.label} digest`,
 		summary:
-			markdown.replaceAll(/\s+/g, " ").trim().slice(0, 280) ||
+			normalized.slice(0, 280) ||
+			neutralFallback ||
 			"No model summary was returned.",
 		keyTopics: [],
 		notableLinks: [],
@@ -968,6 +1019,7 @@ function fallbackDigest(
 function parseDigestFromHybridText(
 	context: PeriodDigestContext,
 	rawText: string,
+	language?: string,
 ): { digest: PeriodDigest; markdown: string } {
 	const [markdownPart, jsonPart] = rawText.split(DELIMITER_PATTERN);
 	const markdown = (markdownPart ?? rawText).trim();
@@ -982,10 +1034,13 @@ function parseDigestFromHybridText(
 				digest: PeriodDigestSchema.parse(JSON.parse(candidate)),
 			};
 		} catch {
-			return { markdown, digest: fallbackDigest(context, markdown) };
+			return {
+				markdown,
+				digest: fallbackDigest(context, markdown, language),
+			};
 		}
 	}
-	return { markdown, digest: fallbackDigest(context, markdown) };
+	return { markdown, digest: fallbackDigest(context, markdown, language) };
 }
 
 function emitVisibleDelta(
@@ -1129,7 +1184,9 @@ function createOpenAIRequestBody(
 			},
 			{
 				role: "user",
-				content: buildPrompt(context),
+				content: buildPrompt(context, {
+					language: languageFromOptions(options),
+				}),
 			},
 		],
 	};
@@ -1172,7 +1229,11 @@ function readOpenAIStreamEffect(
 			}
 
 			const parsed = yield* tryDigestSync(() =>
-				parseDigestFromHybridText(context, state.rawText),
+				parseDigestFromHybridText(
+					context,
+					state.rawText,
+					languageFromOptions(options),
+				),
 			);
 			const cacheKey = digestCacheKey(context, options);
 			const updatedAt = yield* tryDigestSync(() =>
@@ -1213,16 +1274,20 @@ export function streamPeriodDigestEffect(
 	handlers: PeriodDigestStreamHandlers = {},
 ): Effect.Effect<PeriodDigestRunResult, Error> {
 	return Effect.gen(function* () {
+		const resolvedOptions = {
+			...options,
+			language: yield* tryDigestSync(() => languageFromOptions(options)),
+		};
 		yield* refreshPeriodDigestInputsEffect(
-			options,
+			resolvedOptions,
 			{ threads: false },
 			handlers,
 		).pipe(Effect.catchAll(() => Effect.void));
 		let context = yield* tryDigestSync(() =>
-			collectPeriodDigestContext(options),
+			collectPeriodDigestContext(resolvedOptions),
 		);
-		let cacheKey = digestCacheKey(context, options);
-		const cached = options.refresh
+		let cacheKey = digestCacheKey(context, resolvedOptions);
+		const cached = resolvedOptions.refresh
 			? null
 			: yield* tryDigestSync(() =>
 					readSyncCache<{
@@ -1253,7 +1318,7 @@ export function streamPeriodDigestEffect(
 		}
 
 		yield* refreshPeriodDigestInputsEffect(
-			options,
+			resolvedOptions,
 			{
 				timeline: false,
 				mentions: false,
@@ -1264,8 +1329,10 @@ export function streamPeriodDigestEffect(
 			},
 			handlers,
 		).pipe(Effect.catchAll(() => Effect.void));
-		context = yield* tryDigestSync(() => collectPeriodDigestContext(options));
-		cacheKey = digestCacheKey(context, options);
+		context = yield* tryDigestSync(() =>
+			collectPeriodDigestContext(resolvedOptions),
+		);
+		cacheKey = digestCacheKey(context, resolvedOptions);
 
 		const apiKey = process.env.OPENAI_API_KEY;
 		if (!apiKey) {
@@ -1277,12 +1344,12 @@ export function streamPeriodDigestEffect(
 		const response = yield* tryDigestPromise(() =>
 			fetch("https://api.openai.com/v1/responses", {
 				method: "POST",
-				signal: options.signal,
+				signal: resolvedOptions.signal,
 				headers: {
 					authorization: `Bearer ${apiKey}`,
 					"content-type": "application/json",
 				},
-				body: JSON.stringify(createOpenAIRequestBody(context, options)),
+				body: JSON.stringify(createOpenAIRequestBody(context, resolvedOptions)),
 			}),
 		);
 		if (!response.ok) {
@@ -1296,7 +1363,12 @@ export function streamPeriodDigestEffect(
 				),
 			);
 		}
-		return yield* readOpenAIStreamEffect(response, context, options, handlers);
+		return yield* readOpenAIStreamEffect(
+			response,
+			context,
+			resolvedOptions,
+			handlers,
+		);
 	});
 }
 
@@ -1310,6 +1382,9 @@ export function streamPeriodDigest(
 export const __test__ = {
 	PeriodDigestSchema,
 	buildPrompt,
+	digestCacheKey,
+	languageFromOptions,
+	normalizeDigestLanguage,
 	readOpenAIStreamEffect,
 	parseDigestFromHybridText,
 	processSseChunk,

package/src/routes/api/period-digest.tsx

@@ -3,11 +3,13 @@ import { Effect } from "effect";
 import { maybeAutoUpdateBackupEffect } from "#/lib/backup";
 import { runEffectBackground } from "#/lib/effect-runtime";
 import {
+	jsonResponse,
 	parseBoundedInteger,
 	runRouteEffect,
 	sensitiveRequestErrorResponse,
 } from "#/lib/http-effect";
 import {
+	normalizeDigestLanguage,
 	streamPeriodDigestEffect,
 	type PeriodDigestOptions,
 	type PeriodDigestStreamEvent,
@@ -28,6 +30,9 @@ function parseOptions(url: URL): PeriodDigestOptions {
 		includeDms: parseBoolean(url.searchParams.get("includeDms")),
 		refresh: parseBoolean(url.searchParams.get("refresh")),
 		model: url.searchParams.get("model") === "gpt-5.5" ? "gpt-5.5" : undefined,
+		language: normalizeDigestLanguage(
+			url.searchParams.get("language") ?? undefined,
+		),
 		maxTweets: parseBoundedInteger(url.searchParams.get("maxTweets"), {
 			max: 5_000,
 		}),
@@ -60,9 +65,20 @@ export const Route = createFileRoute("/api/period-digest")({
 						const denied = sensitiveRequestErrorResponse(request);
 						if (denied) return denied;
 
-						yield* maybeAutoUpdateBackupEffect();
 						const url = new URL(request.url);
-						const options = parseOptions(url);
+						let options: PeriodDigestOptions;
+						try {
+							options = parseOptions(url);
+						} catch (error) {
+							return jsonResponse(
+								{
+									ok: false,
+									error: error instanceof Error ? error.message : String(error),
+								},
+								{ status: 400 },
+							);
+						}
+						yield* maybeAutoUpdateBackupEffect();
 						let abortDigest: (() => void) | undefined;
 
 						return new Response(

package/src/routes/data-sources.tsx

@@ -169,7 +169,9 @@ function CapabilityRow({
 }
 
 function DataSourcesRoute() {
-	const [snapshot, setSnapshot] = useState<LiveDataSourcesResponse | null>(null);
+	const [snapshot, setSnapshot] = useState<LiveDataSourcesResponse | null>(
+		null,
+	);
 	const [error, setError] = useState<string | null>(null);
 	const [loading, setLoading] = useState(true);
 	const sourcesByKind = useMemo(