Initial release — OpenMontage: the first open-source agentic video production system

11 production pipelines, 47 tools, 124 agent skills. Supports cloud APIs (fal.ai, OpenAI, ElevenLabs, Suno, HeyGen, Runway) and free local providers (diffusers, Piper TTS, WAN 2.1, Hunyuan, CogVideo). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-29 08:25:17 -07:00
commit a3e735cc7a
1147 changed files with 240221 additions and 0 deletions
@@ -0,0 +1,187 @@
+name: talking-head
+version: "2.0"
+description: >
+  End-to-end talking-head video pipeline. Takes raw footage of a person speaking,
+  transcribes it, makes edit decisions, generates subtitles, mixes audio,
+  and composes the final output.
+category: talking_head
+stability: beta
+default_checkpoint_policy: guided
+
+required_skills:
+  - pipelines/talking-head/executive-producer
+  - pipelines/talking-head/idea-director
+  - pipelines/talking-head/script-director
+  - pipelines/talking-head/scene-director
+  - pipelines/talking-head/asset-director
+  - pipelines/talking-head/edit-director
+  - pipelines/talking-head/compose-director
+  - pipelines/talking-head/publish-director
+  - meta/reviewer
+  - meta/checkpoint-protocol
+
+orchestration:
+  mode: executive-producer
+  skill: pipelines/talking-head/executive-producer
+  budget_default_usd: 0.50
+  max_revisions_per_stage: 3
+  max_send_backs: 3
+
+compatible_playbooks:
+  - clean-professional
+
+stages:
+  - name: idea
+    skill: pipelines/talking-head/idea-director
+    produces:
+      - brief
+    tools_available: []
+    checkpoint_required: true
+    human_approval_default: true
+    review_focus:
+      - Brief accurately reflects footage content
+      - Target platform is appropriate
+      - Duration target is realistic
+    success_criteria:
+      - Schema-valid brief artifact
+      - Brief references the provided raw footage
+
+  - name: script
+    skill: pipelines/talking-head/script-director
+    required_artifacts_in:
+      - brief
+    produces:
+      - script
+    required_tools:
+      - transcriber      # speech-to-text
+    optional_tools: []
+    tools_available:
+      - transcriber
+    checkpoint_required: true
+    human_approval_default: true
+    review_focus:
+      - Transcription accuracy (spot-check key sections)
+      - Section boundaries align with topic changes
+      - Timestamps match actual speech
+    success_criteria:
+      - Schema-valid script artifact
+      - Script sections have valid timestamps from transcript data
+
+  - name: scene_plan
+    skill: pipelines/talking-head/scene-director
+    required_artifacts_in:
+      - script
+    optional_artifacts_in:
+      - brief
+    produces:
+      - scene_plan
+    required_tools: []
+    optional_tools:
+      - frame_sampler    # representative frame extraction
+    tools_available:
+      - frame_sampler
+    checkpoint_required: true
+    human_approval_default: true
+    review_focus:
+      - Scenes cover full footage duration
+      - Enhancement opportunities identified
+      - Overlay scenes are feasible
+    success_criteria:
+      - Schema-valid scene_plan artifact
+      - Scenes cover the full duration of the raw footage
+
+  - name: assets
+    skill: pipelines/talking-head/asset-director
+    required_artifacts_in:
+      - scene_plan
+    optional_artifacts_in:
+      - script
+    produces:
+      - asset_manifest
+    required_tools:
+      - subtitle_gen     # captions
+    optional_tools:
+      - audio_mixer      # audio prep
+      - image_selector
+    tools_available:
+      - subtitle_gen
+      - audio_mixer
+      - image_selector
+    checkpoint_required: true
+    human_approval_default: false
+    review_focus:
+      - Subtitle file exists and matches transcript timing
+      - Audio extracted and normalized
+      - All asset paths resolve to existing files
+    success_criteria:
+      - Schema-valid asset_manifest artifact
+      - All referenced asset files exist on disk
+
+  - name: edit
+    skill: pipelines/talking-head/edit-director
+    required_artifacts_in:
+      - scene_plan
+      - asset_manifest
+    optional_artifacts_in:
+      - script
+    produces:
+      - edit_decisions
+    tools_available: []
+    checkpoint_required: true
+    human_approval_default: false
+    review_focus:
+      - Cuts reference valid source files
+      - Subtitles enabled with proper styling
+      - Audio configuration is complete
+    success_criteria:
+      - Schema-valid edit_decisions artifact
+      - All cuts reference valid source files
+      - Timestamps are within source duration bounds
+
+  - name: compose
+    skill: pipelines/talking-head/compose-director
+    required_artifacts_in:
+      - edit_decisions
+      - asset_manifest
+    optional_artifacts_in:
+      - scene_plan
+    produces:
+      - render_report
+    required_tools:
+      - video_compose    # rendering
+      - audio_mixer      # audio layering
+    optional_tools:
+      - face_enhance     # face enhancement
+      - color_grade      # color correction
+      - audio_enhance    # audio cleanup
+    tools_available:
+      - video_compose
+      - audio_mixer
+      - face_enhance
+      - color_grade
+      - audio_enhance
+    checkpoint_required: true
+    human_approval_default: false
+    review_focus:
+      - Output video exists and is playable
+      - Subtitles are synced with speech
+      - Audio is clear with balanced levels
+    success_criteria:
+      - Schema-valid render_report artifact
+      - Output video file exists and is playable
+
+  - name: publish
+    skill: pipelines/talking-head/publish-director
+    required_artifacts_in:
+      - render_report
+    produces:
+      - publish_log
+    tools_available: []
+    checkpoint_required: true
+    human_approval_default: true
+    review_focus:
+      - Metadata is complete and compelling
+      - Export package is structured correctly
+    success_criteria:
+      - Schema-valid publish_log artifact
+      - Export package or publish URLs are present