
from look-tongji-notes10
CLI-first skill to log into Tongji Look, transcribe lecture videos to SRT/TXT, download slide snapshots, and generate Markdown study notes and a timeline outlin
Provides a CLI workflow for interacting with the Tongji Look platform: store IAM credentials locally, list courses, transcribe lecture videos to SRT and TXT, download slide snapshots, and produce student-ready Markdown study notes and a timeline outline derived from transcript + slides. The skill separates responsibilities: the bundled CLI handles authentication, downloading, and ASR; the agent reads the generated artifacts and writes the final Markdown notes and timeline.
Use this skill when you need to convert a Tongji Look lecture into searchable transcripts, time-aligned outlines, and polished study notes—for example, after recording a lecture, when preparing revision materials, or when archiving course content. It is CLI-forward and intended for terminal-based automation or scripts.
Designed for agents that can run local CLIs and read file outputs (Claude Code, Copilot/Codex-style integrations, and assistant runtimes that support filesystem/script execution).
A CLI-first skill for Tongji University's Look platform that logs in, lists courses, transcribes lectures to SRT/TXT, downloads slide snapshots, and generates Markdown study notes. Two scripts: a large main CLI (look_tongji.py) and a timeline formatting utility (timeline_tools.py). Neither ran successfully — the main script depends on an unbundled 'tongji_backend' package, and timeline_tools.py requires subcommand arguments. The SKILL.md is thorough with clear trigger phrases, output contracts, and reference docs. Security is solid: credentials stored in .env (gitignored), no exfiltration risks, no destructive commands.
tongji_backendWell-structured niche skill for Chinese university lecture note generation. Clean security posture. Main dependency (tongji_backend) is external and not included, making out-of-the-box usage require setup. timeline_tools.py is a solid utility with good input validation. The parallel transcript+slide workflow in cmd_note is well-designed.