browser-use/video-use

This system is a video processing tool designed to automate post-production tasks such as filler word removal, auto color grading, and subtitle burning. It operates on a 'read, don't watch' architecture that leverages text transcripts and on-demand visual composites to generate final videos. The system includes robust helpers for handling orientation metadata, HDR tone-mapping, and multi-track transcription, ensuring output correctness through a self-evaluation loop.

The video processing helpers now include an automatic color-grading mode that analyzes clip brightness and saturation to apply subtle, bounded corrections by default, while also adding presets like 'subtle' and 'neutral\_punch'. Rendering fixes ensure portrait videos are detected correctly by accounting for rotation metadata, and HDR sources (HLG/PQ) are automatically tone-mapped to Rec.709 SDR to prevent oversaturation on social platforms. Subtitle vertical margins have been increased to 90 to keep captions clear of UI overlays on vertical video formats. Additionally, the transcription helpers now support selecting specific audio tracks via a new --audio-track argument and refuse to upload silent audio tracks to avoid wasted API calls.

helpers · high confidence

The project has been renamed from 'video-editor' to 'video-use' and is now released under the MIT License. The repository now includes a dedicated \install.md\ file that provides a structured, agent-driven setup process for integrating the skill into various AI agents (such as Claude Code, Codex, and Hermes), covering dependency installation, skill registration, and API key configuration. The \README.md\ has been significantly expanded to detail the product's capabilities, including filler word removal, auto color grading, subtitle burning, and parallel animation generation, while also explaining the core 'read, don't watch' architecture that uses text transcripts and on-demand visual composites. The \SKILL.md\ has been updated to reflect the new name, introduce 'Hard Rules' for production correctness, and describe the self-evaluation loop that verifies rendered output before user presentation. Additionally, a \poster.html\ file was added to visualize the tool's internal timeline and transcript processing.

(repo-wide) · high confidence

Added test coverage for the video rendering helper module, specifically validating the new frame-rate parsing and source-probing logic (including fallbacks to 24 fps) and the updated portrait detection that now relies on stream side-data rotation metadata rather than legacy tags.

tests · high confidence

The package name has been changed from 'video-editor' to 'video-use', and an MIT license file has been linked in the project metadata.

(dependencies) · high confidence

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

• First survey — no prior run to compare against. CAI 49.

• Code Health 96
• Architecture 100
• Maturity 63
• Readiness 33
• Security 89
• Accessibility 43

• 21 commits — 15 feature/other, 6 fixes

• (root) — 11 commits
• helpers/render.py — 5 commits
• (repo) — 2 commits
• helpers/pack_transcripts.py — 1 commit
• helpers/transcribe.py — 1 commit
• skills/manim-video — 1 commit

• fix: fix(pack_transcripts): write output as UTF-8 to avoid cp1252 encoding errors (#10)
• fix: fix(render): account for rotation metadata in portrait detection (#137)
• fix: fix(render): preserve orientation for portrait video sources (#29)
• fix: fix(render): preserve source frame rate by default; add --fps override (#55)
• fix: fix(render): tone-map HLG/PQ sources to Rec.709 SDR (#6)
• fix: fix(subtitles): raise MarginV to 90 to clear vertical safe zones (#5)
• change: Add Browser Use Box editing note
• change: Add MIT license (#32)
• change: Merge pull request #34 from browser-use/add-bux-video-use-note
• change: Merge pull request #91 from ShawnPana/patch-1
• change: Update README.md
• change: add install.md and one-paste setup prompt (#9)
• change: add what-it-does section to readme
• change: autograde + autosound
• change: docs: add HyperFrames animation option (#13)
• change: poster
• change: rename to video-use + readme hero
• change: self eval loop
• change: transcribe: select the audio track, and refuse to upload a silent one (#134)
• change: trim readme — get started first, tighten how-it-works
• …and 1 more

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

browser-use/video-use was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.