@clipform/mcp-server 1.44.3 → 1.44.4

package/README.md

@@ -85,7 +85,7 @@ Your MCP client lists these automatically on connect (via `tools/list`). Full re
 | `clipform_search_music` | Search for royalty-free music tracks and ambient sounds. |
 | `clipform_list_compositions` | Browse available video compositions and their expected props schemas. |
 | `clipform_list_assets` | List available creative assets (sound effects, animations, fonts) for video compositions. |
-| `clipform_check_render` | Check the status of a render job started by clipform_generate_video or clipform_render_composition. |
+| `clipform_check_render` | Check the status of render jobs started by clipform_generate_video or clipform_render_composition. |
 | `clipform_fetch_boundary` | Fetch a GeoJSON boundary polygon for a country, city, or region. |
 | `clipform_get_guide` | Retrieve craft knowledge for building a specific form type. |
 | `clipform_get_workflow` | Retrieve a step-by-step build workflow for creating a specific form type. |

package/dist/{chunk-VTRO7ZC3.js → chunk-6IRFVCIG.js}

@@ -4124,7 +4124,7 @@ function quizWorkflow(args, sessionContext, authMode = "authenticated") {
 4. **Generate narration** with clipform_generate_tts - batch ALL questions in a single call (pass an array of items). Tease the question - do NOT reveal the answer or read options aloud. Keep each narration 5-15 seconds.
 5. **Build video** for each question - pick per question; mixing treatments across one quiz is fine:
    - Default - narrated slideshow: clipform_search_media (kind: "image", 3 images per question, batch all queries in one call), then clipform_generate_video with wait: false per question - fires all renders in parallel - then collect URLs with clipform_check_render.
-   - Designed visual - when a composition fits the question or the user asks for one (e.g. a grid reveal): clipform_list_compositions to see what's available, then clipform_render_composition with wait: false per clip. Compositions take narration via sound.narration.
+   - Designed visual - when a composition fits the question or the user asks for one (e.g. a grid reveal): clipform_list_compositions to see what's available, then ONE clipform_render_composition call with items (all clips render in parallel), collected with one clipform_check_render (job_ids). Compositions take narration via sound.narration.
 6. ${ATTACH_MEDIA_STEP}
 7. ${REPUBLISH_STEP}
 8. **Log** with clipform_log_generation (sources, images, attributions)
@@ -4226,7 +4226,7 @@ function compositionQuizWorkflow(args, sessionContext, authMode = "authenticated
 
 1. **Discover what's renderable** - call clipform_list_compositions. Availability varies by session; choose mechanics ONLY from what it returns. Match each question to a mechanic via the composition guide's matrix. Mix 2-3 mechanics across the quiz; for questions with no fitting composition, use the base quiz's Ken Burns pipeline instead.
 2. **Write questions** - difficulty curve as in the base guide. Difficulty = subject obscurity (Japan's flag vs Kiribati's), never clue degradation. ${count} questions \u2248 ${count * 2 + 1} nodes (each question is a clue + reveal pair) - check the plan's node limit.
-3. **Render all clips** with clipform_render_composition, wait: false on EVERY call - fire the full batch (clue clip + reveal clip per question), then collect URLs with clipform_check_render. Chain-state compositions: clue clip animate: true, reveal clip animate: false + answer props.
+3. **Render all clips** in ONE clipform_render_composition call with items - the full batch (clue clip + reveal clip per question) renders in parallel - then collect URLs with one clipform_check_render (job_ids). Chain-state compositions: clue clip animate: true, reveal clip animate: false + answer props.
 4. ${BUILD_ONCE_STEP}
    - Settings: show_step_counter: true, disable_back_navigation: true
    - Per question, a PAIR of nodes:
@@ -4709,4 +4709,4 @@ export {
   getWorkflowText,
   registerPrompts
 };
-//# sourceMappingURL=chunk-VTRO7ZC3.js.map
+//# sourceMappingURL=chunk-6IRFVCIG.js.map