Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/sonichi/sutando/media-transcriptnpx skills add sonichi/sutando --skill media-transcriptgit clone --depth 1 https://github.com/sonichi/sutandoWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.00525 |
| Opus 5 | $0.00000 | $0.00262 |
| Sonnet 5 | $0.00000 | $0.00105 |
| Haiku 4.5 | $0.00000 | $0.00052 |
Grade A, and why
media-transcript scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 30 lines — stays where its author put it; the contents beside it link to each section on GitHub.
media-transcript
Pull a video's transcript (captions/subtitles) as clean text — the doc-ingest sibling for video. Any task that references a YouTube (or other yt-dlp-supported) URL can consume the spoken content as text instead of being blind to it.
Usage:
python3 skills/media-transcript/scripts/transcript.py <url> [--lang en] [--timestamps] [--json]
Prints the transcript to stdout. --timestamps prefixes blocks with [mm:ss] cue starts (use when the question asks when/after what something was said). --json wraps the result as {"url", "ok", "kind", "text"|"error"} for programmatic callers.
When to use
- A task references a video URL whose content matters ("what did they say about X", "what number is mentioned after Y", summarizing a talk).
- The agent-eval harness hits video-based benchmark tasks (4 of 10 GAIA L3 fails on 2026-07-30 were video-content questions this skill unblocks).
- Anything needing spoken-word content where captions exist.
Not for:
- Local audio/video files — use
skills/audio-transcribe(whisper on the audio track); the script points there and exits 3. - Videos with no captions at all — v1 is captions-only (uploader subs preferred, auto-captions fallback). The script says so honestly (exit 1) rather than guessing; a whisper-on-downloaded-audio fallback is a possible v2, deliberately out of scope (media downloads are heavyweight and often unnecessary).
Behavior
- Prefers uploader subtitles over auto-generated captions;
--langpreference (defaulten) with graceful fallback to available English variants. - Downloads ONLY the subtitle track — never the media stream — into a private temp dir, removed afterwards. No API keys.
- Auto-caption VTTs repeat rolling caption windows; the parser dedupes them, strips inline tags/entities, and drops cue numbers, producing readable prose.
yt-dlpis probed at runtime (PATH binary, else the importable module) — a host without it gets one actionable error naming the install, never a crash (same design language as doc-ingest's extractor probing).
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 30 lines · 0 tokens per session scan A 73971d86292b
media-transcript is a skill published in the GitHub repository sonichi/sutando (389 stars, last pushed 2d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 525 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
ask
Ask the user one or more multiple-choice questions with an interactive in-app card — pause for their decisions, then continue with the options they pick or the custom instructions they type. Also the delivery mechanism for quizzes and knowledge checks.
knowledge
Maintain your own long-term memory — record a new belief, correct a wrong fact, unlearn a lesson, edit your playbook, identity and standing instructions.
gmail
Manage Gmail email — drafting, sending, organizing, filters, vacation replies, and inbox analysis.
app-builder
Build and edit small, personal visual tools and artifacts — dashboards, trackers, calculators, data visualizations, charts, simple landing pages, and slide decks the user wants for THEMSELVES. This is the right skill whenever the user asks to "visualize this," "make a chart," or "build an artifact" for their own use…
assistant-migration
Migrate from ChatGPT, Claude, OpenClaw, Hermes, Manus, and other AI assistants into Vellum by inspecting their data exports, conversation archives, files, prompts, custom instructions, memory, saved memories, tools, GPTs, workflows, integrations, and relationships, then mapping as much as safely possible into Vellum…
guardian-verify-setup
Set up channel verification for phone, Telegram, Slack, Discord, or email channels via outbound verification flow.