pull - #490
Open
RiCSaucd wants to merge 10 commits into
Open
Conversation
provider_menu() previously called get_info() and get_status() for each tool, doubling dependency checks. Reuse the status already produced by get_info(), with a fallback when status is malformed, and add a regression test to lock in single status evaluation. Co-authored-by: Copilot App <[email protected]>
…enMontage pipeline (#1) * feat: why-the-sky-is-blue explainer — research through edit checkpoints Co-authored-by: RiCSaucd <[email protected]> * feat: why-the-sky-is-blue compose checkpoint + Piper sentence-silence quirk skill Co-authored-by: RiCSaucd <[email protected]> * fix: address PR #1 review feedback on checkpoint metadata - Set pipeline_type to animated-explainer across all stage checkpoints - Point Piper quirk references to skills/core/piper-tts-quirks.md - Correct s4 review_notes durations and spacing in checkpoint_assets - Align compose total_duration_seconds with rendered MP4 (62.55s) - Embed final_review artifact and fix render_report audit refs Co-authored-by: RiCSaucd <[email protected]> * fix: address second round of PR #1 review feedback - Align d-008 sample-gate rationale with actual s4 duration (-2.5%) - Fix narration-s1 review_notes to reference s4 gate consistently - Correct d-010 skill paths to skills/core/piper-tts-quirks.md - Embed transcript_comparison results (150/150 words, accuracy 1.0) - Document data_explainer motion_ratio=0.0 as valid per delivery_promise Co-authored-by: RiCSaucd <[email protected]> * fix: address CodeRabbit PR review on sky video checkpoints - Count img-sunset-gradient in images_generated; clarify provider vs procedural - Align cut-4 stat to 5.9x (blue vs red) matching bar chart; update script/scene_plan - Set subtitle_check subtitles_expected/present true in compose final_review - Use canonical assets/ paths for narration, music, and sunset background in compose Co-authored-by: RiCSaucd <[email protected]> * fix: align caption stat text and proposal refs with review - Patch compose caption 'nine' -> 'six' to match script/provider_text - Update proposal checkpoint stat references from 9.4x to 5.9x blue-vs-red Co-authored-by: RiCSaucd <[email protected]> --------- Co-authored-by: Cursor Agent <[email protected]> Co-authored-by: RiCSaucd <[email protected]>
* feat: why-the-sky-is-blue explainer — research through edit checkpoints Co-authored-by: RiCSaucd <[email protected]> * feat: why-the-sky-is-blue compose checkpoint + Piper sentence-silence quirk skill Co-authored-by: RiCSaucd <[email protected]> * fix: address PR #1 review feedback on checkpoint metadata - Set pipeline_type to animated-explainer across all stage checkpoints - Point Piper quirk references to skills/core/piper-tts-quirks.md - Correct s4 review_notes durations and spacing in checkpoint_assets - Align compose total_duration_seconds with rendered MP4 (62.55s) - Embed final_review artifact and fix render_report audit refs Co-authored-by: RiCSaucd <[email protected]> * fix: address second round of PR #1 review feedback - Align d-008 sample-gate rationale with actual s4 duration (-2.5%) - Fix narration-s1 review_notes to reference s4 gate consistently - Correct d-010 skill paths to skills/core/piper-tts-quirks.md - Embed transcript_comparison results (150/150 words, accuracy 1.0) - Document data_explainer motion_ratio=0.0 as valid per delivery_promise Co-authored-by: RiCSaucd <[email protected]> * feat: produce 30s viral Netflix Documentary coffee vertical video - Netflix Documentary TikTok trend parody: Chapter One: The Pour - 1080x1920 vertical render via Remotion (explainer-data, tiktok profile) - Free stack: Piper TTS, Pixabay cinematic music, word-level captions - Production runner at scripts/produce_viral_netflix_coffee.py - Output: projects/viral-netflix-coffee/renders/chapter-one-the-pour.mp4 (~30s) Co-authored-by: RiCSaucd <[email protected]> * fix: address PR #15 review feedback on viral video checkpoints - Rewrite decision_log.json to schema-valid decisions[] format - Fix cut timing via concat section boundaries (not broken phrase search) - Transcribe with faster-whisper vad_filter=false (Piper speech was dropped) - Populate composition.captions and align render_report duration with ffprobe - Add decision_log_ref, final_review_ref, audio_codec, file_size_bytes - Use relative paths in final_review; document edit vs encoded duration padding - Reuse cached Pixabay music when search returns HTTP 403 Co-authored-by: RiCSaucd <[email protected]> * fix: address CodeRabbit PR review on sky video checkpoints - Count img-sunset-gradient in images_generated; clarify provider vs procedural - Align cut-4 stat to 5.9x (blue vs red) matching bar chart; update script/scene_plan - Set subtitle_check subtitles_expected/present true in compose final_review - Use canonical assets/ paths for narration, music, and sunset background in compose Co-authored-by: RiCSaucd <[email protected]> * fix: align caption stat text and canonical asset paths with review - Patch sky compose caption 'nine' -> 'six' to match script/provider_text - Update proposal checkpoint stat references from 9.4x to 5.9x blue-vs-red - Use assets/narration and assets/music paths in viral compose checkpoint - Persist canonical audio paths in produce script checkpoint output Co-authored-by: RiCSaucd <[email protected]> * feat: integrate Cinematic Intro Engine as cinematic-intro pipeline Add the 7-stage Seedance 2.0 intro workflow (line sheet → character sheet → beat map → storyboard → board gate → test/final render → VO delivery) as a first-class OpenMontage pipeline with director skills, prompt library HTML, routing in AGENT_GUIDE/INDEX, and contract tests. Co-authored-by: RiCSaucd <[email protected]> --------- Co-authored-by: Cursor Agent <[email protected]> Co-authored-by: RiCSaucd <[email protected]>
* feat: why-the-sky-is-blue explainer — research through edit checkpoints Co-authored-by: RiCSaucd <[email protected]> * feat: why-the-sky-is-blue compose checkpoint + Piper sentence-silence quirk skill Co-authored-by: RiCSaucd <[email protected]> * fix: address PR #1 review feedback on checkpoint metadata - Set pipeline_type to animated-explainer across all stage checkpoints - Point Piper quirk references to skills/core/piper-tts-quirks.md - Correct s4 review_notes durations and spacing in checkpoint_assets - Align compose total_duration_seconds with rendered MP4 (62.55s) - Embed final_review artifact and fix render_report audit refs Co-authored-by: RiCSaucd <[email protected]> * fix: address second round of PR #1 review feedback - Align d-008 sample-gate rationale with actual s4 duration (-2.5%) - Fix narration-s1 review_notes to reference s4 gate consistently - Correct d-010 skill paths to skills/core/piper-tts-quirks.md - Embed transcript_comparison results (150/150 words, accuracy 1.0) - Document data_explainer motion_ratio=0.0 as valid per delivery_promise Co-authored-by: RiCSaucd <[email protected]> * feat: produce 30s viral Netflix Documentary coffee vertical video - Netflix Documentary TikTok trend parody: Chapter One: The Pour - 1080x1920 vertical render via Remotion (explainer-data, tiktok profile) - Free stack: Piper TTS, Pixabay cinematic music, word-level captions - Production runner at scripts/produce_viral_netflix_coffee.py - Output: projects/viral-netflix-coffee/renders/chapter-one-the-pour.mp4 (~30s) Co-authored-by: RiCSaucd <[email protected]> * fix: address PR #15 review feedback on viral video checkpoints - Rewrite decision_log.json to schema-valid decisions[] format - Fix cut timing via concat section boundaries (not broken phrase search) - Transcribe with faster-whisper vad_filter=false (Piper speech was dropped) - Populate composition.captions and align render_report duration with ffprobe - Add decision_log_ref, final_review_ref, audio_codec, file_size_bytes - Use relative paths in final_review; document edit vs encoded duration padding - Reuse cached Pixabay music when search returns HTTP 403 Co-authored-by: RiCSaucd <[email protected]> * fix: address CodeRabbit PR review on sky video checkpoints - Count img-sunset-gradient in images_generated; clarify provider vs procedural - Align cut-4 stat to 5.9x (blue vs red) matching bar chart; update script/scene_plan - Set subtitle_check subtitles_expected/present true in compose final_review - Use canonical assets/ paths for narration, music, and sunset background in compose Co-authored-by: RiCSaucd <[email protected]> * fix: align caption stat text and canonical asset paths with review - Patch sky compose caption 'nine' -> 'six' to match script/provider_text - Update proposal checkpoint stat references from 9.4x to 5.9x blue-vs-red - Use assets/narration and assets/music paths in viral compose checkpoint - Persist canonical audio paths in produce script checkpoint output Co-authored-by: RiCSaucd <[email protected]> --------- Co-authored-by: Cursor Agent <[email protected]> Co-authored-by: RiCSaucd <[email protected]>
There was a problem hiding this comment.
Pull request overview
This PR expands OpenMontage’s video-editing and ingest capabilities by adding a Descript-style text-based editor (CutScript OSS integration), enhancing video_downloader for production ingest + limited playlist support, and introducing a new “cinematic-intro” pipeline with associated skills/contracts and agent-definition validation.
Changes:
- Add
text_based_editortool + supporting transcript segment utilities and CutScript-derived FFmpeg helpers (vendored with license/notice). - Extend
video_downloaderwithingest_modedefaults (reference vs production) and optional bounded playlist download, plus contract tests. - Add “cinematic-intro” pipeline definition + stage-director skills, and add
.claude/agentsschema/loader + contract tests.
Reviewed changes
Copilot reviewed 53 out of 54 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| tools/video/text_based_editor.py | New transcript-driven editing tool (compute segments, export, studio sound). |
| tools/tool_registry.py | Reuse tool status from get_info() when possible. |
| tools/analysis/video_downloader.py | Add ingest modes, playlist support, normalized metadata, and audio extraction helper. |
| third_party/cutscript/video_editor.py | Vendored FFmpeg cutting/export helpers (stream-copy + re-encode). |
| third_party/cutscript/audio_cleaner.py | Vendored noise reduction helper (DeepFilterNet or FFmpeg fallback). |
| third_party/cutscript/init.py | Package init for CutScript-derived utilities. |
| third_party/cutscript/NOTICE.md | Attribution notice for vendored CutScript code. |
| third_party/cutscript/LICENSE | MIT license for CutScript vendored files. |
| third_party/init.py | Add third_party package marker. |
| lib/text_edit/segments.py | Transcript word normalization + keep-segment computation + filler detection utilities. |
| tests/contracts/test_text_based_editor.py | Contract tests for text_based_editor and segment utilities. |
| tests/contracts/test_video_downloader.py | Contract tests for new video_downloader ingest/playlist behavior. |
| tests/contracts/test_phase0_contracts.py | Add coverage ensuring provider menu doesn’t redundantly call get_status(). |
| requirements.txt | Add yt-dlp dependency range for video_downloader. |
| .agents/skills/text-based-editing/SKILL.md | New Layer-3 skill doc for transcript-driven editing workflows. |
| .agents/skills/video-download/SKILL.md | Document OpenMontage video_downloader parameters and playlist behavior. |
| pipeline_defs/talking-head.yaml | Make text_based_editor available for talking-head pipeline edit stage. |
| pipeline_defs/clip-factory.yaml | Make text_based_editor optional/available in clip-factory edit stage. |
| pipeline_defs/cinematic-intro.yaml | New cinematic-intro pipeline manifest (stages, gates, required skills/tools). |
| skills/creative/video-editing.md | Add text_based_editor to editing tool routing list. |
| skills/creative/cinematic-intro-engine.md | New “Cinematic Intro Engine” routing/prompt library doc. |
| skills/pipelines/cinematic-intro/executive-producer.md | New stage orchestration guidance for cinematic-intro pipeline. |
| skills/pipelines/cinematic-intro/proposal-director.md | New proposal stage director for cinematic-intro. |
| skills/pipelines/cinematic-intro/script-director.md | New script stage director for cinematic-intro. |
| skills/pipelines/cinematic-intro/scene-director.md | New scene_plan stage director for cinematic-intro. |
| skills/pipelines/cinematic-intro/edit-director.md | New edit stage director for cinematic-intro. |
| skills/pipelines/cinematic-intro/asset-director.md | New assets stage director for cinematic-intro. |
| skills/pipelines/cinematic-intro/compose-director.md | New compose stage director for cinematic-intro. |
| skills/pipelines/cinematic-intro/publish-director.md | New publish stage director for cinematic-intro. |
| skills/INDEX.md | Index updates: add cinematic-intro pipeline skills + Piper quirks entry. |
| skills/core/piper-tts-quirks.md | Document Piper CLI corruption quirk and workaround. |
| lib/agent_loader.py | Add loader/validator for .claude/agents/*.md frontmatter contract. |
| schemas/agents/agent_definition.schema.json | JSON schema for agent frontmatter validation. |
| tests/lib/test_agent_loader.py | Unit tests for lib/agent_loader.py. |
| tests/contracts/test_agent_definition_contracts.py | Contract tests to enforce agent definition validity in CI. |
| .claude/agents/sre.md | Add SRE subagent definition. |
| .claude/agents/image-prompt-engineer.md | Add image prompt engineering subagent definition. |
| .claude/agents/ai-data-remediation-engineer.md | Add AI data remediation subagent definition. |
| skills/pipelines/cinematic-intro/* | Stage-specific runbooks for cinematic-intro pipeline. |
| tests/contracts/test_cinematic_intro_pipeline.py | Contract test for cinematic-intro pipeline manifest/skills presence. |
| scripts/produce_viral_netflix_coffee.py | Add one-shot production script (viral-netflix-coffee). |
| pipelines/why-the-sky-is-blue/decision_log.json | Add decision log artifact for sample pipeline run. |
| pipelines/why-the-sky-is-blue/checkpoint_*.json | Add checkpoint artifacts for sample pipeline run. |
| pipelines/viral-netflix-coffee/decision_log.json | Add decision log artifact for sample pipeline run. |
| pipelines/viral-netflix-coffee/checkpoint_compose.json | Add compose checkpoint artifact for sample pipeline run. |
| AGENTS.md | Expand notes; reiterate AGENT_GUIDE as source of truth plus env notes. |
| AGENT_GUIDE.md | Add cinematic-intro to pipeline table + routing hint. |
| .devcontainer/devcontainer.json | Add minimal devcontainer config. |
Suppressed comments (1)
tools/video/text_based_editor.py:305
output_pathis aPath, so theif output_path else ""branch never selects the empty-string default (becausePath('')becomesPath('.'), which is truthy). This makesstudio_soundbehave incorrectly whenoutput_pathis omitted, and can also let a missing/emptyinput_pathslip through (sincePath('.')exists).
Prefer validating the raw strings, ensure input_path is a file, and only pass an explicit output_path to clean_audio when provided.
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Comment on lines
+240
to
+245
| input_path = Path(inputs.get("input_path", "")) | ||
| output_path = Path(inputs.get("output_path", "")) | ||
| if not input_path.exists(): | ||
| return ToolResult(success=False, error=f"Input not found: {input_path}") | ||
| if not output_path: | ||
| return ToolResult(success=False, error="output_path is required for export") |
Comment on lines
+164
to
+172
| idempotency_key_fields = [ | ||
| "url", | ||
| "format", | ||
| "max_resolution", | ||
| "ingest_mode", | ||
| "allow_playlist", | ||
| "max_playlist_items", | ||
| "max_duration_seconds", | ||
| ] |
Cloud Agent HTTPS currently allowlists GitHub only. Provider keys in .env cannot call api.x.ai (or other Cloudflare-fronted APIs) until those hosts are allowlisted. Co-authored-by: Cursor Agent <[email protected]> Co-authored-by: RiCSaucd <[email protected]>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Related issue
Closes #
Changes
Testing
Checklist
make test-contracts/make test) where applicable.