fix(cli): keep --slides value from eating input and validate it

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## 0.23.1 - Unreleased
 
+- CLI slides: preserve URLs and media inputs following bare `--slides`, reject malformed per-run slide options, and return HTTP 400 for invalid daemon requests while preserving stored-config tolerance (#488, thanks @vincent-peng).
 - CLI completions: offer every shipped interface locale across commands, keep locale values separate from daemon subcommands, and restore missing existing flag suggestions and documentation (#486, thanks @vincent-peng).
 
 ## 0.23.0 - 2026-09-21

docs/commands/summarize.md

@@ -87,9 +87,9 @@ If `[input]` is omitted, summarize prints concise help and exits.
 
 ### Slides
 
-`--slides [value]` / `--no-slides`
+`--slides [on|off]` / `--no-slides`
 : Extract slides from a YouTube URL, direct video URL, or local video file and render inline alongside the summary. Combine with `--extract` to interleave slides in the full transcript. See [Slides mode](../slides.md).
-: Use `--no-slides` to disable slide extraction enabled in the config file for one run.
+: The optional value accepts boolean words (`on`/`off`, `true`/`false`, `yes`/`no`, `1`/`0`); anything else is rejected. Use `--no-slides` to disable slide extraction enabled in the config file for one run.
 
 `--slides-ocr` / `--no-slides-ocr`
 : Run OCR on extracted slides. Requires `tesseract`.

src/daemon/server-summarize-request.ts

@@ -35,12 +35,9 @@ function resolveRequestSlidesSettings({
     explicitEnvKey?: string,
   ) => string | null;
 }): SlideSettings | null {
-  const slidesValue = request.slides;
-  const tesseractAvailable = resolveToolPath("tesseract", env, "TESSERACT_PATH") !== null;
-  const slidesOcrValue = tesseractAvailable ? request.slidesOcr : false;
-  return resolveSlideSettings({
-    slides: slidesValue,
-    slidesOcr: slidesOcrValue,
+  const input = {
+    slides: request.slides,
+    slidesOcr: request.slidesOcr,
     // Daemon/API callers may be browser-extension or localhost clients that
     // only need to request extraction, not select host filesystem paths. Keep
     // slide artifacts under the per-user Summarize directory so an authenticated
@@ -51,7 +48,11 @@ function resolveRequestSlidesSettings({
     slidesMax: request.slidesMax,
     slidesMinDuration: request.slidesMinDuration,
     cwd: resolveHomeDir(env),
-  });
+  };
+  const settings = resolveSlideSettings(input);
+  if (!settings?.ocr || resolveToolPath("tesseract", env, "TESSERACT_PATH") !== null)
+    return settings;
+  return resolveSlideSettings({ ...input, slidesOcr: false });
 }
 
 export type ParsedSummarizeRequest = {
@@ -131,7 +132,6 @@ export async function parseSummarizeRequest({
   const format: "text" | "markdown" =
     formatRaw === "markdown" || formatRaw === "md" ? "markdown" : "text";
   const overrides = resolveRunOverrides(obj);
-  const slidesSettings = resolveRequestSlidesSettings({ env, request: obj, resolveToolPath });
   const diagnostics = parseDiagnostics(obj.diagnostics);
   const hasText = Boolean(textContent.trim());
 
@@ -150,6 +150,15 @@ export async function parseSummarizeRequest({
     return null;
   }
 
+  let slidesSettings: SlideSettings | null;
+  try {
+    slidesSettings = resolveRequestSlidesSettings({ env, request: obj, resolveToolPath });
+  } catch (error) {
+    const message = error instanceof Error ? error.message : String(error);
+    json(res, 400, { ok: false, error: message }, cors);
+    return null;
+  }
+
   return {
     pageUrl,
     title,

src/run/help.ts

@@ -92,7 +92,7 @@ export function buildProgram(locale: CliLocale = "en") {
         .default("auto"),
     )
     .option(
-      "--slides [value]",
+      "--slides [on|off]",
       t(
         "extract.slides.for.youtube.direct.video.urls.or.local.video.files.and.render.them.inline.inside.the.summary.narrative.when.supported.combine.with.extract.to.interleave.slides.in.the.full.transcript",
       ),

src/run/runner-setup.ts

@@ -10,7 +10,7 @@ export function prepareRunEnvironment(
   argv: string[],
   inputEnv: Record<string, string | undefined>,
 ) {
-  const normalizedArgv = normalizeDiarizeArgv(argv);
+  const normalizedArgv = normalizeMediaInputArgv(argv);
   const preSeparatorArgv = argvBeforeSeparator(normalizedArgv);
   const noColorFlag = preSeparatorArgv.includes("--no-color");
   let envForRun: Record<string, string | undefined> = noColorFlag
@@ -52,15 +52,15 @@ export function stripCliLocaleArgs(argv: readonly string[], locale: CliLocale =
   return stripped;
 }
 
-export function normalizeDiarizeArgv(argv: string[]): string[] {
+export function normalizeMediaInputArgv(argv: string[]): string[] {
   const separatorIndex = argv.indexOf("--");
   return argv.map((arg, index) => {
     if (separatorIndex !== -1 && index > separatorIndex) return arg;
-    if (arg !== "--diarize") return arg;
+    const value = arg === "--diarize" ? "auto" : arg === "--slides" ? "true" : null;
+    if (value === null) return arg;
     const next = argv[index + 1];
     if (!next || next.startsWith("-")) return arg;
-    if (["auto", "elevenlabs", "openai"].includes(next.toLowerCase())) return arg;
-    return /^[a-z][a-z\d+.-]*:\/\//i.test(next) || isDirectMediaUrl(next) ? "--diarize=auto" : arg;
+    return /^[a-z][a-z\d+.-]*:\/\//i.test(next) || isDirectMediaUrl(next) ? `${arg}=${value}` : arg;
   });
 }
 

src/slides/settings.ts

@@ -28,20 +28,24 @@ const DEFAULT_MIN_DURATION_SECONDS = 2;
 const DECIMAL_INT_PATTERN = /^\d+$/;
 const DECIMAL_NUMBER_PATTERN = /^\d+(?:\.\d+)?$/;
 
-const parseBoolean = (raw: unknown): boolean | null => {
+const parseBoolean = (raw: unknown, label: string): boolean | null => {
+  if (raw == null) return null;
   if (typeof raw === "boolean") return raw;
-  if (typeof raw !== "string") return null;
-  const normalized = raw.trim().toLowerCase();
-  if (!normalized) return null;
-  if (["1", "true", "yes", "on"].includes(normalized)) return true;
-  if (["0", "false", "no", "off"].includes(normalized)) return false;
-  return null;
+  if (typeof raw === "string") {
+    const normalized = raw.trim().toLowerCase();
+    if (["1", "true", "yes", "on"].includes(normalized)) return true;
+    if (["0", "false", "no", "off"].includes(normalized)) return false;
+  }
+  throw new Error(`Unsupported ${label}: ${String(raw)}`);
 };
 
 const parsePositiveInt = (raw: unknown, label: string, min = 1): number | null => {
   if (raw == null) return null;
   const value = typeof raw === "string" ? raw.trim() : raw;
-  if (typeof value === "string" && !DECIMAL_INT_PATTERN.test(value)) {
+  if (
+    typeof value !== "number" &&
+    (typeof value !== "string" || !DECIMAL_INT_PATTERN.test(value))
+  ) {
     throw new Error(`Unsupported ${label}: ${String(raw)}`);
   }
   const numeric = typeof value === "number" ? value : Number(value);
@@ -61,7 +65,10 @@ const parseNumberInRange = (
 ): number | null => {
   if (raw == null) return null;
   const value = typeof raw === "string" ? raw.trim() : raw;
-  if (typeof value === "string" && !DECIMAL_NUMBER_PATTERN.test(value)) {
+  if (
+    typeof value !== "number" &&
+    (typeof value !== "string" || !DECIMAL_NUMBER_PATTERN.test(value))
+  ) {
     throw new Error(`Unsupported ${label}: ${String(raw)}`);
   }
   const numeric = typeof value === "number" ? value : Number(value);
@@ -75,13 +82,9 @@ const parseNumberInRange = (
 };
 
 export function resolveSlideSettings(input: SlideSettingsInput): SlideSettings | null {
-  const slidesFlag = parseBoolean(input.slides);
-  const ocrFlag = parseBoolean(input.slidesOcr);
+  const slidesFlag = parseBoolean(input.slides, "--slides");
+  const ocrFlag = parseBoolean(input.slidesOcr, "--slides-ocr");
   const enabled = Boolean((slidesFlag ?? false) || (ocrFlag ?? false));
-  if (!enabled) return null;
-
-  const dirRaw = typeof input.slidesDir === "string" ? input.slidesDir.trim() : DEFAULT_OUTPUT_DIR;
-  const outputDir = path.resolve(input.cwd, dirRaw || DEFAULT_OUTPUT_DIR);
 
   const sceneThreshold =
     parseNumberInRange(input.slidesSceneThreshold, "--slides-scene-threshold", {
@@ -94,6 +97,10 @@ export function resolveSlideSettings(input: SlideSettingsInput): SlideSettings |
       min: 0,
       max: 86_400,
     }) ?? DEFAULT_MIN_DURATION_SECONDS;
+  if (!enabled) return null;
+
+  const dirRaw = typeof input.slidesDir === "string" ? input.slidesDir.trim() : DEFAULT_OUTPUT_DIR;
+  const outputDir = path.resolve(input.cwd, dirRaw || DEFAULT_OUTPUT_DIR);
   return {
     enabled,
     ocr: Boolean(ocrFlag ?? false),

tests/cli.flags.test.ts

@@ -17,7 +17,8 @@ import {
 } from "../src/flags.js";
 import { buildProgram } from "../src/run/help.js";
 import { resolveRunnerFlags } from "../src/run/runner-flags.js";
-import { normalizeDiarizeArgv, prepareRunEnvironment } from "../src/run/runner-setup.js";
+import { normalizeMediaInputArgv, prepareRunEnvironment } from "../src/run/runner-setup.js";
+import { resolveSlideSettings } from "../src/slides/index.js";
 
 describe("cli flag parsing", () => {
   it("defaults summary length to long", () => {
@@ -36,7 +37,7 @@ describe("cli flag parsing", () => {
 
   it("treats a URL after bare --diarize as the positional input", () => {
     const url = "https://www.youtube.com/watch?v=abcdefghijk";
-    const argv = normalizeDiarizeArgv(["--diarize", url]);
+    const argv = normalizeMediaInputArgv(["--diarize", url]);
     const program = buildProgram();
     program.parse(argv, { from: "user" });
 
@@ -48,7 +49,7 @@ describe("cli flag parsing", () => {
   it.each(["recording.mp3", "/tmp/interview.mp4"])(
     "treats %s after bare --diarize as the positional input",
     (input) => {
-      const argv = normalizeDiarizeArgv(["--diarize", input]);
+      const argv = normalizeMediaInputArgv(["--diarize", input]);
       const program = buildProgram();
       program.parse(argv, { from: "user" });
 
@@ -60,15 +61,82 @@ describe("cli flag parsing", () => {
 
   it("keeps explicit diarization providers intact", () => {
     const url = "https://www.youtube.com/watch?v=abcdefghijk";
-    expect(normalizeDiarizeArgv(["--diarize", "openai", url])).toEqual([
+    expect(normalizeMediaInputArgv(["--diarize", "openai", url])).toEqual([
       "--diarize",
       "openai",
       url,
     ]);
   });
 
   it("keeps bare --diarize unchanged when no positional input follows", () => {
-    expect(normalizeDiarizeArgv(["--diarize"])).toEqual(["--diarize"]);
+    expect(normalizeMediaInputArgv(["--diarize"])).toEqual(["--diarize"]);
+  });
+
+  it("treats a URL after bare --slides as the positional input", () => {
+    const url = "https://www.youtube.com/watch?v=abcdefghijk";
+    const { normalizedArgv: argv } = prepareRunEnvironment(["--slides", url], {});
+    const program = buildProgram();
+    program.parse(argv, { from: "user" });
+
+    expect(argv).toEqual(["--slides=true", url]);
+    expect(program.opts().slides).toBe("true");
+    expect(program.args).toEqual([url]);
+  });
+
+  it.each(["lecture.mp4", "/tmp/talk.mkv"])(
+    "treats %s after bare --slides as the positional input",
+    (input) => {
+      const argv = normalizeMediaInputArgv(["--slides", input]);
+      const program = buildProgram();
+      program.parse(argv, { from: "user" });
+
+      expect(argv).toEqual(["--slides=true", input]);
+      expect(program.opts().slides).toBe("true");
+      expect(program.args).toEqual([input]);
+    },
+  );
+
+  it("keeps explicit boolean values after --slides intact", () => {
+    const url = "https://example.com";
+    expect(normalizeMediaInputArgv(["--slides", "off", url])).toEqual(["--slides", "off", url]);
+    const program = buildProgram();
+    program.parse(["--slides", "off", url], { from: "user" });
+    expect(program.opts().slides).toBe("off");
+    expect(program.args).toEqual([url]);
+  });
+
+  it("keeps bare --slides unchanged when no positional input follows", () => {
+    expect(normalizeMediaInputArgv(["--slides"])).toEqual(["--slides"]);
+    expect(normalizeMediaInputArgv(["--slides", "--json"])).toEqual(["--slides", "--json"]);
+  });
+
+  it("keeps bare --slides before the -- separator unchanged", () => {
+    expect(normalizeMediaInputArgv(["--slides", "--", "clip.mp4"])).toEqual([
+      "--slides",
+      "--",
+      "clip.mp4",
+    ]);
+    expect(normalizeMediaInputArgv(["--", "--slides", "clip.mp4"])).toEqual([
+      "--",
+      "--slides",
+      "clip.mp4",
+    ]);
+  });
+
+  it("rejects unsupported --slides and --slides-ocr values", () => {
+    expect(() => resolveSlideSettings({ slides: "bogus", cwd: "/" })).toThrow(
+      /Unsupported --slides: bogus/,
+    );
+    expect(() => resolveSlideSettings({ slides: 1, cwd: "/" })).toThrow(/Unsupported --slides: 1/);
+    expect(() => resolveSlideSettings({ slides: true, slidesOcr: "bogus", cwd: "/" })).toThrow(
+      /Unsupported --slides-ocr: bogus/,
+    );
+  });
+
+  it("accepts boolean words for --slides and rejects explicit empty values", () => {
+    expect(resolveSlideSettings({ slides: "yes", cwd: "/" })?.enabled).toBe(true);
+    expect(resolveSlideSettings({ slides: "off", cwd: "/" })).toBeNull();
+    expect(() => resolveSlideSettings({ slides: "", cwd: "/" })).toThrow(/Unsupported --slides:/);
   });
 
   it("parses speaker identity profiles and repeatable timestamp anchors", () => {

tests/daemon.request-slides-settings.test.ts

@@ -93,4 +93,44 @@ describe("parseSummarizeRequest slides settings", () => {
       sceneThreshold: 0.5,
     });
   });
+
+  it.each([
+    [{ slides: "" }, /Unsupported --slides:/],
+    [{ slidesOcr: " " }, /Unsupported --slides-ocr:/],
+    [{ slidesMax: "bogus" }, /Unsupported --slides-max: bogus/],
+    [{ slides: false, slidesSceneThreshold: 2 }, /Unsupported --slides-scene-threshold: 2/],
+    [{ slidesMax: true }, /Unsupported --slides-max: true/],
+    [{ slidesSceneThreshold: [0.5] }, /Unsupported --slides-scene-threshold: 0.5/],
+    [{ slidesMinDuration: false }, /Unsupported --slides-min-duration: false/],
+    [{ slides: "bogus" }, /Unsupported --slides: bogus/],
+    [{ slides: 1 }, /Unsupported --slides: 1/],
+    [{ slidesOcr: "bogus" }, /Unsupported --slides-ocr: bogus/],
+    [{ slides: true, slidesMax: "bogus" }, /Unsupported --slides-max: bogus/],
+    [{ slides: true, slidesSceneThreshold: 2 }, /Unsupported --slides-scene-threshold: 2/],
+    [{ slides: true, slidesMinDuration: -1 }, /Unsupported --slides-min-duration: -1/],
+  ])("rejects invalid slide option %o with a 400 response", async (fields, errorPattern) => {
+    for (const toolsAvailable of [true, false]) {
+      const writeHead = vi.fn();
+      const end = vi.fn();
+      const res = { writeHead, end } as unknown as http.ServerResponse;
+
+      const parsed = await parseSummarizeRequest({
+        req: createJsonRequest({
+          url: "https://example.com/video.mp4",
+          mode: "url",
+          ...fields,
+        }),
+        res,
+        cors: {},
+        env: { HOME: "/home/alice" },
+        resolveToolPath: toolsAvailable ? resolveToolPath : () => null,
+      });
+
+      expect(parsed).toBeNull();
+      expect(writeHead).toHaveBeenCalledWith(400, expect.any(Object));
+      const body = JSON.parse(String(end.mock.calls[0]?.[0])) as { ok?: boolean; error?: string };
+      expect(body.ok).toBe(false);
+      expect(body.error).toMatch(errorPattern);
+    }
+  });
 });