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 skills add brian-huanggg/youtube-mcp --skill youtube-watchgit clone --depth 1 https://github.com/brian-huanggg/youtube-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/brian-huanggg/youtube-mcp/youtube-watch)<a href="https://agentmods.dev/skills/brian-huanggg/youtube-mcp/youtube-watch"><img src="https://agentmods.dev/badge/skills/brian-huanggg/youtube-mcp/youtube-watch/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/brian-huanggg/youtube-mcp/youtube-watch"><img src="https://agentmods.dev/badge/skills/brian-huanggg/youtube-mcp/youtube-watch.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00081 | $0.00687 |
| Opus 5 | $0.00041 | $0.00344 |
| Sonnet 5 | $0.00016 | $0.00137 |
| Haiku 4.5 | $0.00008 | $0.00069 |
Grade A, and why
youtube-watch 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 11d 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 — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Watching a YouTube video
Goal: get a video's content into context efficiently — never dump a whole transcript when a search or a slice will do. Long videos otherwise blow the token budget.
Decide the approach with get_info first
Call get_info(url) (MCP) or youtube-watch-mcp-cli info URL. It returns title, duration,
chapters, and has_captions — cheap, no transcript. Use it to choose:
- Short video (< ~10 min) →
get_transcript, then read the whole file. - Long video →
get_transcriptfor the path + preview, thensearch_transcriptorget_segmentto pull only what's relevant. Do not read the full file. - No captions (
has_captions: false) → transcript needs ASR (--asr, slower, opt-in); warn the user before doing it.
Tools (MCP server youtube-watch-mcp)
| Tool | When |
|---|---|
get_info(url) |
Always first. Metadata + caption availability. |
get_transcript(url, lang?) |
Get the cleaned transcript. Returns a file path + preview, not the full text. |
search_transcript(url, query) |
Find where something is said. Returns timestamped snippets. |
get_segment(url, start, end) |
Read one time range (seconds). |
get_transcript writes the cleaned transcript to a cache file and returns its
path. For a short video, read that file. For a long one, prefer search/segment.
Reading the transcript file
The path from get_transcript points at cleaned prose in
~/.cache/youtube-mcp/<video_id>/transcript.txt. Read it with the normal Read
tool, or grep it — it's already de-duplicated and stripped of markup/noise.
CLI fallback (no MCP server)
If the MCP server isn't connected, use the command directly:
youtube-watch-mcp-cli info "URL"
youtube-watch-mcp-cli transcript "URL" # -> prints cache path + preview
youtube-watch-mcp-cli search "URL" "query" # -> [m:ss] snippet
youtube-watch-mcp-cli segment "URL" 120 180 # -> text in [120s, 180s)
Answering the user
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.
- 11d ago First seen · 58 lines · 81 tokens per session scan A 39d8e726d020
youtube-watch is a skill published in the GitHub repository brian-huanggg/youtube-mcp (0 stars, last pushed 2mo ago), licensed MIT. It adds 81 tokens to every session and 687 once invoked, about $0.0004 per session on Opus 5. 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 2026-08-31.
Other skills, from other repositories
narsil
Use narsil-mcp code intelligence tools effectively. Use when searching code, finding symbols, analyzing call graphs, scanning for security vulnerabilities, exploring dependencies, or performing static analysis on indexed repositories.
cel-lib-doc-generator
Generates and updates standardized README.md documentation for CEL extension libraries in C++, Go, and Java by extracting declarations, functions, macros, types, signatures, and examples from source code.
cel-debugging
Diagnosing and resolving Common Expression Language (CEL) compilation and evaluation errors.
media-download
Download video (or audio) from YouTube, Rumble, and other yt-dlp-supported sites via the media-downloader MCP server, and natively store the downloaded media into the epistemic-graph knowledge graph as a content-addressed blob + :MediaAsset. Use when the agent must fetch a video by URL and optionally persist the raw…
media-audio
Extract audio as MP3 from a video/media URL via the media-downloader MCP server, natively storing the resulting audio into epistemic-graph as a content-addressed blob + :MediaAsset. Use when the agent needs the audio track (podcast, music, lecture) rather than the full video. Do NOT use for full-video download…
cel-authoring
Authoring, configuring, and testing Common Expression Language (CEL) expressions, policies, rules, and environment JSON configurations.