ryanshi1103
UserSafety-bounded AI skills for review-first media workflows
Categories
Indexed Skills (10)
confera-audio-cleanup
Plan conservative, local audio cleanup through registered denoise capabilities. Use when an agent should recommend noise reduction for approved media without constructing shell commands, overwriting source audio, or starting a render.
confera-caption-layout
Recommend readable caption layout from supplied caption segments. Use when an agent should propose line breaks, timing presentation, or placement without changing transcript review state or embedding subtitles into a final export.
confera-chatcut-exchange
Prepare or validate a consented manual ChatCut exchange package. Use when an agent should describe a privacy-limited proxy, subtitle, storyboard, or timeline handoff without logging in, automating a private web service, uploading files, or approving imported edits.
confera-ffmpeg-render
Translate reviewed edit intent into registered FFmpeg capability choices. Use when an agent should produce a non-executable render plan while a trusted application layer retains control of commands, paths, validation, finalization, and export approval.
confera-media-inspector
Create evidence-bound media inspection plans from controlled metadata. Use when an agent should select a registered ffprobe capability without reading arbitrary files, constructing commands, uploading original media, or making unsupported content claims.
confera-narration-draft
Draft narration candidates that remain linked to reviewed evidence. Use when an agent should propose voice-over text without inventing facts, calling text-to-speech, imitating a person's voice, or marking narration reviewed.
confera-photo-polish
Recommend conservative, non-destructive photo corrections from approved metrics or keyframes. Use when an agent should suggest exposure, color, denoise, or sharpening changes without altering faces, inferring identity or emotion, accessing originals, or executing an edit.
confera-quality-review
Produce a read-only final media quality report from controlled proxy evidence and metadata. Use when an agent should identify technical risks without modifying media, approving a revision, finalizing artifacts, or starting export.
confera-story-editor
Propose evidence-bound storyboard revisions from reviewed excerpts and controlled artifact metadata. Use when an agent should suggest narrative structure without reading arbitrary files, mutating the current revision, approving the storyboard, or exporting media.
confera-timeline-review
Review a media timeline and propose a new candidate revision. Use when an agent should identify pacing, ordering, overlap, or duration issues without mutating the current timeline, preserving an old export approval, approving changes, or rendering media.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.