This skill provides planning and guidance for KrillinAI's end-to-end pipeline: downloading media, transcribing, translating, generating TTS dubs, and rendering final videos in landscape or portrait formats. It's aimed at orchestrating multi-stage jobs (subtitle, tts, horizontal/vertical renders) and contains verification and recovery advice for interrupted runs.
Use this skill when you need a full localization workflow executed or planned: e.g., translating and dubbing a YouTube video into multiple languages, producing portrait versions for TikTok, or running a multi-output render that includes subtitles, dubbing, and covers. If pipeline execution isn't wired in for your environment, follow the recommended per-stage commands.
Best used with CLI-capable agent tooling that can run shell commands (Copilot/Codex-style or CLI-focused agents) and agents that understand multi-step media workflows. The repo and docs indicate human+agent orchestration models rather than a web UI.
KrillinAI Pipeline is a planning skill for multi-stage video localization workflows (subtitle, TTS, render) using the KrillinAI CLI. The SKILL.md itself acknowledges that pipeline execution may not be wired in yet, directing users to prefer individual commands. No scripts were bundled, so only static analysis was performed. The skill is narrowly scoped to one proprietary CLI tool and its practical value is limited until pipeline execution is actually implemented.
No security concerns — no scripts, no network calls, no credentials. The skill is essentially a reference card for CLI output options. Low code quality because the instructions are minimal and the skill admits its own primary command may not work. Architecture is adequate with proper frontmatter and a reference to a contract doc, but no scripts/ directory or references/ separation visible.