AskVideo — YouTube Transcripts & AI Video Chat Skill

AskVideo — YouTube Transcripts & AI Video Chat

Two core capabilities for agents:

Transcripts — pull the full transcript (with timestamps) of any YouTube video in one command. Fast, no indexing needed.
AI chat — ask questions, get summaries, extract information from a video using AI (RAG over the transcript).

When to Use This Skill

Use transcript when:

User asks for the transcript, subtitles, or captions of a YouTube video
User wants the raw text of a video to paste somewhere, translate, or process further
You (the agent) need to read a video's contents to reason about it
User asks for specific quotes or wants to search the video text

Use ask / chat when:

User asks a question about a video ("what does he say about X?")
User wants a summary, bullet points, key takeaways
User asks for timestamps where a topic is discussed
User wants structured extraction (list of tools, steps, code snippets, etc.)

Keywords: YouTube transcript, video transcript, video captions, YouTube subtitles, video summary, explain video, what's in this video, video content, video tutorial, video Q&A

Prerequisites

The askvideo CLI must be installed and authenticated:

# Check if installed
which askvideo

# Install if missing
npm install -g askvideo

# First time — create an account (email OTP, free)
askvideo signup

# Or, if the user already has an account
askvideo login

# Sign out when done (optional)
askvideo logout

Signup and login both use email OTP — the user enters their email, receives a 6-digit code, and is authenticated on this machine. The API key is stored in the OS config dir; subsequent commands are non-interactive.

Commands

Get a Transcript (Recommended for Agents)

transcript (alias yt) returns the full transcript of a video. No indexing, no AI — fast.

# Plain transcript with timestamps (default)
askvideo transcript "<youtube-url>"

# Short alias
askvideo yt "<youtube-url>"

# Plain text, no timestamps
askvideo transcript "<youtube-url>" --no-timestamps

# JSON output (best for parsing in agents)
askvideo transcript "<youtube-url>" --json

# Save to file
askvideo transcript "<youtube-url>" -o transcript.txt
askvideo transcript "<youtube-url>" -f srt -o subtitles.srt

# Quiet mode (just the transcript, no decorative output)
askvideo transcript "<youtube-url>" --quiet

Output formats:

txt (default) — readable text, optionally with [m:ss] timestamps
srt — SubRip subtitle format, suitable for video players
json — { success, video_id, transcript, segments[], duration }

Ask a Single Question

ask indexes the video (first call only) and answers a question. Use --simple for clean non-streaming output.

askvideo ask "<question>" --url "<youtube-url>" --simple

# Examples
askvideo ask "Summarize this video in 3 bullet points" --url "<url>" --simple
askvideo ask "At what timestamp do they discuss pricing?" --url "<url>" --simple
askvideo ask "List all the tools mentioned in this video" --url "<url>" --simple

Interactive Chat

askvideo chat "<youtube-url>"

List Indexed Videos

askvideo videos

Check Credits

askvideo credits

Workflow for Agents

Fast path: transcript only

If the user asks for the transcript or you just need the text, use transcript:

askvideo transcript "<url>" --json --quiet

Parse the JSON, use transcript or segments[] as needed.

AI path: question answering

If the user asks a question about the video, use ask:

askvideo ask "<user's question>" --url "<url>" --simple

First call indexes the video (~30-60s). Subsequent calls with --id <yt_id> are instant.

Auth prerequisite

Before any other command, ensure the user is authenticated. If they see Authentication required, have them run:

askvideo signup   # new user
askvideo login    # existing user

Response Format

Always present the answer clearly:

Based on the video:

[output from askvideo]

For transcripts, consider summarizing or trimming if the transcript is very long before showing to the user.

Error Handling

| Error | Solution | |-------|----------| | Authentication required | Run askvideo signup (new user) or askvideo login (existing) | | No video credits remaining | User needs to upgrade at https://askvideo.ai/pricing | | No message credits remaining | User needs to upgrade at https://askvideo.ai/pricing | | Failed to fetch transcript | Video has no captions — try a different video | | Invalid YouTube URL | Check the URL format |

Supported Video Types

Standard YouTube videos with captions/auto-generated transcripts
YouTube Shorts
Unlisted videos (if URL is provided)

Limitations

Requires the video to have a transcript (auto-generated or manual)
transcript command is generous; no per-call charges at current free tier
ask/chat use video credits: free plan gets 1 video + 10 messages/month
Video indexing (first ask call) takes 30-60 seconds

Examples for Common Tasks

Get a clean transcript as plain text

askvideo transcript "<url>" --no-timestamps --quiet

Get a transcript as JSON for programmatic use

askvideo transcript "<url>" --json

Save as SRT subtitles

askvideo yt "<url>" -f srt -o video.srt

Summarize a video

askvideo ask "Give me a comprehensive summary with key takeaways" --url "<url>" --simple

Extract code/commands

askvideo ask "List all the code snippets or commands mentioned" --url "<url>" --simple

Find a specific topic with timestamp

askvideo ask "When and what does the speaker say about <topic>?" --url "<url>" --simple

Agent Skills: AskVideo — YouTube Transcripts & AI Video Chat

Install this agent skill to your local

Skill Files