@pixeldojo/mcp 0.25.0 → 0.26.0

package/README.md

@@ -3,7 +3,7 @@
 PixelDojo MCP server. One install for Claude Code, Cursor, OpenClaw, and any
 other [Model Context Protocol](https://modelcontextprotocol.io) host.
 
-23 named tools, 140+ models, async by default.
+25 named tools, 140+ models, async by default.
 
 **Prefer zero install?** PixelDojo also runs a hosted MCP server with OAuth
 sign-in (no API key): point any remote-capable client at
@@ -58,6 +58,7 @@ Restart your agent. You'll see 23 new tools.
 | `pixeldojo_overlay` | Ad text layouts on a video: hook title, lower-third callout, review card, progress bar, countdown timer, price tag, offer badge and a branded end card with logo. Exact text, your brand color, clear of the platform UI. |
 | `pixeldojo_caption` | Animated word-by-word captions on a video with speech, placed clear of the TikTok/Reels/Shorts buttons. Pop, Highlighter or Clean, any highlight color. |
 | `pixeldojo_audio` | Full audio scenes: dialogue + effects + music in one clip (Seed Audio 1.0). Reference clips are hosted URLs; local audio files are not accepted. |
+| `pixeldojo_effects` / `pixeldojo_effect` | One-click effects with the prompt built in: list them (283 video effects like AI Hug or Earth Zoom In, 30 photo effects like action figure or Ghibli style, and one-job tools like watermark or background removal), then run one by slug on your photos. |
 | `pixeldojo_library` | Search your media library: saved Library items + recent generations. |
 | `pixeldojo_save_workflow` / `run_workflow` / `list_workflows` | Save a multi-step chain once, run it by name from any chat. |
 | `pixeldojo_upload` | Local file → 24h temp URL, for reference inputs. |

package/dist/cli.cjs

@@ -1839,14 +1839,17 @@ async function plan(input, ctx) {
     look: input.look ?? (music ? MUSIC_VIDEO_LOOKS[0] : "live-action"),
     lengthTarget: input.length,
     ...!music && input.pacing === "trailer" ? { pacing: "trailer" } : {},
+    ...input.quality === "final" ? { clipQuality: "final" } : {},
     assetIds: [],
     ...music ? {
       // No script step names a music video, so it would stay "Untitled film".
       ...input.idea?.trim() ? { title: input.idea.trim().slice(0, 80) } : {},
       musicVideoMode: true,
       musicUrl: input.song_url,
-      ...input.song_duration_sec ? { musicDurationSec: input.song_duration_sec } : {},
-      clipEngine: MUSIC_CLIP_ENGINE_ID
+      ...input.song_duration_sec ? { musicDurationSec: input.song_duration_sec } : {}
+      // Engine left on Auto, like the web composer's music mode, so the
+      // planner routes a mature song to H3 and the rest to P-Video 2 Pro
+      // Reference (tests/contracts/mcp-film-music-parity.test.ts).
     } : {}
   });
   if (!created.data.film) return { kind: "text", text: `Could not create the film: ${created.data.error ?? created.status}` };
@@ -2248,14 +2251,13 @@ async function run10(input, ctx) {
   if (action === "redo") return redoShots(input, ctx);
   return advance(input.film_id, input.soundtrack, input.music_job_id, ctx);
 }
-var STORY_LOOKS, MUSIC_VIDEO_LOOKS, LOOKS, MUSIC_CLIP_ENGINE_ID, InputSchema10, InputJsonSchema10, shots, benignDispatchRefusal, count, frameDone, RETRY_CLEARS, REFUSED, isRefused, PLANNER_POLL_MS, PLANNER_MAX_POLLS, isMusicVideo, needsSlice, stillPlanningText, LOOK_MUSIC, ARC, filmSkill;
+var STORY_LOOKS, MUSIC_VIDEO_LOOKS, LOOKS, InputSchema10, InputJsonSchema10, shots, benignDispatchRefusal, count, frameDone, RETRY_CLEARS, REFUSED, isRefused, PLANNER_POLL_MS, PLANNER_MAX_POLLS, isMusicVideo, needsSlice, stillPlanningText, LOOK_MUSIC, ARC, filmSkill;
 var init_film = __esm({
   "src/skills/film.ts"() {
     init_upload_helpers();
     STORY_LOOKS = ["live-action", "anime", "3d-animation", "noir", "horror", "sci-fi", "documentary", "western", "romance", "romantic-comedy", "comedy"];
     MUSIC_VIDEO_LOOKS = ["rap-video", "neon-club", "stage-performance", "retro-camcorder", "y2k-pop", "moody-rnb", "rock-grit", "dream-pop", "k-pop"];
     LOOKS = [...STORY_LOOKS, ...MUSIC_VIDEO_LOOKS];
-    MUSIC_CLIP_ENGINE_ID = "minimax-h3-reference";
     InputSchema10 = zod.z.object({
       idea: zod.z.string().max(2e3).optional().describe("One-to-three sentence film idea. The studio writes the screenplay (free). Provide idea OR script for planning."),
       script: zod.z.string().max(2e4).optional().describe("A full screenplay to plan instead of an idea (sluglines, action lines, dialogue)."),
@@ -2266,6 +2268,7 @@ var init_film = __esm({
       lyrics: zod.z.string().max(1e4).optional().describe("Music video mode: the song lyrics, so shots land on the right lines. Optional."),
       aspect: zod.z.enum(["16:9", "9:16"]).default("16:9"),
       pacing: zod.z.enum(["story", "trailer"]).default("story").describe("'story' (default) plays every shot's action whole. 'trailer' cuts it like a trailer or promo: most shots keep only their first 1.5-3 seconds."),
+      quality: zod.z.enum(["standard", "final"]).default("standard").describe("'standard' (default) renders 720p clips and a 720p film. 'final' renders 1080p clips where the engine supports it and exports the film at 1080p; it costs more per second."),
       length: zod.z.enum(["30s", "60s", "2-3min", "5min"]).default("60s").describe("Target length of the finished film."),
       film_id: zod.z.string().max(40).optional().describe("Existing film id (from a previous call) for produce/status/cast/redo."),
       action: zod.z.enum(["plan", "produce", "status", "cast", "redo"]).default("plan").describe("'plan' creates + storyboards (free). 'cast' casts, replaces or removes characters on an existing film. 'produce' starts charged generation. 'status' advances production and reports progress; poll it every ~30s until it returns the finished film. 'redo' renders a new take of the listed shots (charged), then status re-saves and re-exports."),
@@ -2305,6 +2308,11 @@ var init_film = __esm({
         lyrics: { type: "string", description: "Plan step, music video mode (optional): the song lyrics, so shots land on the right lines." },
         aspect: { type: "string", enum: ["16:9", "9:16"], description: "Plan step: frame aspect ratio. Default 16:9." },
         length: { type: "string", enum: ["30s", "60s", "2-3min", "5min"], description: "Plan step: target length. Default 60s. Ignored for a music video, whose length is the song." },
+        quality: {
+          type: "string",
+          enum: ["standard", "final"],
+          description: "Plan step: 'standard' (default) is 720p. 'final' renders 1080p clips where the engine supports it and exports at 1080p, at a higher price per second."
+        },
         pacing: {
           type: "string",
           enum: ["story", "trailer"],
@@ -2853,7 +2861,7 @@ var init_episode = __esm({
   "src/skills/episode.ts"() {
     STYLES2 = ["editorial-motion", "paper-diorama", "hand-drawn", "storybook", "poster-vector", "soft-3d"];
     CATEGORIES = ["education", "history", "kids", "storytelling", "custom"];
-    ENGINES = ["omni-draft", "omni-final", "still-motion"];
+    ENGINES = ["omni-draft", "omni-final", "omni-1080", "still-motion"];
     MIN_SECONDS = 60;
     MAX_SECONDS = 300;
     MAX_IN_FLIGHT = 4;
@@ -2888,7 +2896,7 @@ var init_episode = __esm({
         engine: {
           type: "string",
           enum: [...ENGINES],
-          description: "Create step (optional): render every scene with one engine. omni-draft: fast, cheap motion preview. omni-final: full quality motion with sound. still-motion: one illustration per scene, gently panned, cheapest. Default: the style picks."
+          description: "Create step (optional): render every scene with one engine. omni-draft: fast, cheap motion preview. omni-final: full quality motion with sound (720p). omni-1080: the same at 1080p, and the episode exports at 1080p. still-motion: one illustration per scene, gently panned, cheapest. Default: the style picks."
         },
         music: { type: "boolean", description: "Create step: add a music bed under the narration. Default true." },
         title: { type: "string", description: "Create step (optional): episode title. Defaults to the topic." }
@@ -3581,6 +3589,148 @@ var init_save_character = __esm({
     };
   }
 });
+function priceOf(e) {
+  if (!e.creditsByEngine) return `${e.credits ?? "?"} credits`;
+  return Object.entries(e.creditsByEngine).map(([engine, c]) => `${engine}${engine === e.defaultEngine ? " (default)" : ""} ${c ?? "?"} credits`).join(" / ");
+}
+async function listRun(input, ctx) {
+  const q = new URLSearchParams();
+  for (const [k, v] of Object.entries(input)) if (v !== void 0 && v !== "") q.set(k, String(v));
+  if (!input.limit) q.set("limit", "20");
+  const res = await (ctx.fetch ?? fetch)(`${ctx.baseUrl}/api/v1/effects?${q}`, { headers: headers(ctx, "pixeldojo_effects") });
+  const data = await res.json().catch(() => ({}));
+  if (!res.ok) return { kind: "text", text: `Could not list effects: ${data.error ?? `HTTP ${res.status}`}` };
+  const effects = data.effects ?? [];
+  const cats = Object.entries(data.categories ?? {}).map(([fam, list]) => `${fam}: ${list.join(", ")}`);
+  if (!effects.length) {
+    return { kind: "text", text: ["No effects match. Categories:", ...cats].join("\n") };
+  }
+  const lines = effects.map((e) => {
+    const ins = e.inputs.map(
+      (i) => i.kind === "image" ? `photo "${i.label}"` : `text "${i.label}"${i.options ? ` [${i.options.join(" | ")}]` : ""}${i.optional ? " (optional)" : ""}`
+    );
+    return `- ${e.slug} (${e.family}, ${e.output}, ${priceOf(e)}): ${e.title}. ${e.tagline} Inputs: ${ins.join(", ")}.`;
+  });
+  const shown = data.total && data.total > effects.length ? `, showing ${effects.length}` : "";
+  return {
+    kind: "text",
+    text: [
+      `${data.total ?? effects.length} effects${shown}. Run one with pixeldojo_effect (slug, image_urls in the order listed, texts for text inputs).`,
+      ...lines,
+      ...input.search || input.family || input.category ? [] : ["", "Categories:", ...cats]
+    ].join("\n")
+  };
+}
+async function effectRun(input, ctx) {
+  const images = input.image_paths?.length ? await uploadLocalFiles(input.image_paths, ctx) : input.image_urls ?? [];
+  const res = await (ctx.fetch ?? fetch)(`${ctx.baseUrl}/api/v1/effects/${encodeURIComponent(input.slug)}/run`, {
+    method: "POST",
+    headers: headers(ctx, "pixeldojo_effect"),
+    body: JSON.stringify({
+      images,
+      ...input.texts ? { texts: input.texts } : {},
+      ...input.family ? { family: input.family } : {},
+      ...input.engine ? { engine: input.engine } : {}
+    })
+  });
+  const data = await res.json().catch(() => ({}));
+  if (!res.ok || !data.jobId) {
+    return { kind: "text", text: `Effect not started (nothing charged): ${data.error ?? `HTTP ${res.status}`}` };
+  }
+  const result = await pollJobUntilDoneOrHandoff(data.jobId, (id) => ctx.client.getJob(id), ctx.buildStatusUrl, {
+    now: ctx.pollOverrides?.now,
+    sleep: ctx.pollOverrides?.sleep
+  });
+  return pollResultToSkillResult(
+    result,
+    (urls) => urls.length === 0 ? `Job ${data.jobId} completed but returned no assets.` : `Done (${data.creditCost ?? "?"} credits). Asset: ${urls[0]}`
+  );
+}
+var FAMILIES, ENGINES2, headers, ListSchema, effectsListSkill, RunSchema, runProperties, effectRunSkill;
+var init_effects = __esm({
+  "src/skills/effects.ts"() {
+    init_poll();
+    init_upload_helpers();
+    init_types();
+    FAMILIES = ["video-effect", "photo-effect", "tool"];
+    ENGINES2 = ["klein", "qwen"];
+    headers = (ctx, skill) => ({
+      Authorization: `Bearer ${ctx.apiKey}`,
+      "Content-Type": "application/json",
+      "x-pixeldojo-source": `mcp/${ctx.mcpVersion}`,
+      "x-pixeldojo-skill": skill
+    });
+    ListSchema = zod.z.object({
+      search: zod.z.string().max(200).optional(),
+      family: zod.z.enum(FAMILIES).optional(),
+      category: zod.z.string().max(60).optional(),
+      limit: zod.z.number().int().min(1).max(100).optional()
+    });
+    effectsListSkill = {
+      name: "pixeldojo_effects",
+      description: "List PixelDojo's one-click effects: 283 video effects (a photo or two in, a 5 s clip out: AI Hug, Earth Zoom In, AI Dance), 30 photo effects (a photo in, a styled image out: action figure, Ghibli style, sticker, aging filter) and one-job tools (remove a watermark, object or background, headshots, photo restoration, relight, logo, thumbnail and poster makers). Search by words, family or category; each result gives the slug, inputs and credit price. Free. Run one with pixeldojo_effect.",
+      inputZod: ListSchema,
+      inputJsonSchema: {
+        type: "object",
+        properties: {
+          search: { type: "string", description: 'Words like "hug", "action figure", "ghibli", "remove background".' },
+          family: {
+            type: "string",
+            enum: FAMILIES,
+            description: "video-effect = photo to clip, photo-effect = photo to styled image, tool = one-job edit."
+          },
+          category: { type: "string", description: 'A category name from a previous listing, e.g. "Viral", "Toys & 3D".' },
+          limit: { type: "integer", minimum: 1, maximum: 100, description: "Default 20." }
+        }
+      },
+      allowedModels: [],
+      defaultModel: "",
+      run: listRun
+    };
+    RunSchema = zod.z.object({
+      slug: zod.z.string().min(1).max(80),
+      family: zod.z.enum(FAMILIES).optional(),
+      image_urls: zod.z.array(zod.z.string().url()).max(4).optional(),
+      image_paths: zod.z.array(zod.z.string()).max(4).optional(),
+      texts: zod.z.array(zod.z.string().max(200)).max(4).optional(),
+      engine: zod.z.enum(ENGINES2).optional()
+    });
+    runProperties = {
+      slug: { type: "string", description: 'The effect slug from pixeldojo_effects, e.g. "ai-hug", "action-figure", "watermark-remover".' },
+      family: {
+        type: "string",
+        enum: FAMILIES,
+        description: "Needed only for action-figure, comic-book and red-carpet, which are both a video and a photo effect."
+      },
+      image_urls: {
+        type: "array",
+        items: { type: "string", format: "uri" },
+        description: "The photos as https URLs, in the order the listing names them (most take one; duo effects take two)."
+      },
+      image_paths: {
+        type: "array",
+        items: { type: "string" },
+        description: "Absolute local photo paths instead of image_urls; uploaded first (24h temp storage)."
+      },
+      texts: {
+        type: "array",
+        items: { type: "string" },
+        description: "Text inputs in listing order, for entries that list them (object to remove, new background, brand name)."
+      },
+      engine: { type: "string", enum: ENGINES2, description: "Photo effects only: leave out for the default engine." }
+    };
+    effectRunSkill = {
+      name: "pixeldojo_effect",
+      description: "Run one of PixelDojo's one-click effects on your photos: the prompt is built in, so pass only the slug from pixeldojo_effects and the photos (image_urls, or image_paths for local files), plus texts for tools that ask for them. Video effects return a 5 s clip (usually handed off; poll with pixeldojo_status), photo effects and tools an image. Costs the credits the listing shows.",
+      hostedDescription: "Run one of PixelDojo's one-click effects on your photos: the prompt is built in, so pass only the slug from pixeldojo_effects and the photos as https image_urls, plus texts for tools that ask for them. Video effects return a 5 s clip (usually handed off; poll with pixeldojo_status), photo effects and tools an image. Costs the credits the listing shows.",
+      inputZod: RunSchema,
+      inputJsonSchema: { type: "object", required: ["slug"], properties: runProperties },
+      allowedModels: [],
+      defaultModel: "",
+      run: effectRun
+    };
+  }
+});
 function substitutePlaceholders(args, inputs, prevUrl) {
   const missing = [];
   const fill = (value) => {
@@ -3829,6 +3979,8 @@ __export(skills_exports, {
   captionSkill: () => captionSkill,
   characterSkill: () => characterSkill,
   editSkill: () => editSkill,
+  effectRunSkill: () => effectRunSkill,
+  effectsListSkill: () => effectsListSkill,
   episodeSkill: () => episodeSkill,
   filmSkill: () => filmSkill,
   findHostedSkill: () => findHostedSkill,
@@ -3890,6 +4042,7 @@ var init_skills = __esm({
     init_library();
     init_save();
     init_save_character();
+    init_effects();
     init_workflows();
     ALL_SKILLS = [
       generateSkill,
@@ -3900,6 +4053,8 @@ var init_skills = __esm({
       captionSkill,
       overlaySkill,
       audioSkill,
+      effectsListSkill,
+      effectRunSkill,
       librarySkill,
       saveSkill,
       saveCharacterSkill,
@@ -3941,6 +4096,8 @@ var init_skills = __esm({
       pixeldojo_caption: render("Caption video"),
       pixeldojo_overlay: render("Add offer or end card"),
       pixeldojo_audio: render("Generate music, speech or sound"),
+      pixeldojo_effects: read("List one-click effects"),
+      pixeldojo_effect: render("Run one-click effect"),
       pixeldojo_video: render("Change, animate or extend video"),
       pixeldojo_ad: render("Make video ad"),
       pixeldojo_campaign: render("Make product launch campaign"),
@@ -3963,10 +4120,10 @@ function makeAttributedClient(options) {
   const { mcpVersion, skillName, baseFetch, ...sdkOptions } = options;
   const real = baseFetch ?? globalThis.fetch.bind(globalThis);
   const wrappedFetch = (input, init) => {
-    const headers = new Headers(init?.headers);
-    headers.set("x-pixeldojo-source", `mcp/${mcpVersion}`);
-    if (skillName) headers.set("x-pixeldojo-skill", skillName);
-    return real(input, { ...init, headers });
+    const headers2 = new Headers(init?.headers);
+    headers2.set("x-pixeldojo-source", `mcp/${mcpVersion}`);
+    if (skillName) headers2.set("x-pixeldojo-skill", skillName);
+    return real(input, { ...init, headers: headers2 });
   };
   return new sdk.PixelDojoClient({
     ...sdkOptions,
@@ -4010,7 +4167,7 @@ function makeRedactedLogger(target = console) {
 }
 
 // src/server.ts
-var PACKAGE_VERSION = "0.25.0" ;
+var PACKAGE_VERSION = "0.26.0" ;
 var SERVER_INSTRUCTIONS = "Pixel Dojo generates images, video, audio and short films from one catalog of AI models, billed in credits from the connected Pixel Dojo account. A job that runs past the 30-second wait returns a jobId to check later; check it rather than submitting again, since every new submission is charged. Generated asset URLs are deleted one hour after they are created: pixeldojo_save keeps one in the user's My Media permanently (after a safety check), and finished films stay in the account library. In any prompt, @Name uses a character, product, outfit, location, prop or style the user saved in Character Studio: its photos are attached automatically. pixeldojo_library source=characters lists them, and pixeldojo_save_character saves a new one from 3 to 5 photos.";
 function buildServer(options) {
   if (!options.apiKey) {