Skip to main content
skill CX Codex

video-memory — Claude Code skill: memory of watched videos

A persistent index of every video the agent has analyzed on this machine: frames, OCR, transcripts and past answers. Useful when you need to recall what was in a recording.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add oxbshw/watch-skill --skill video-memory -a claude-code
Open on GitHub

What video-memory does

The skill keeps a library under ~/.watch-skill/. Each analyzed video lands there with an id, title, duration and analysis time, so sessions do not wipe it — the more videos you run through, the denser the index gets.

Search is hybrid: keywords plus semantics, with Arabic folding and CJK and Thai segmentation handled. Hits come back as video plus timestamp, and you can follow up with a question about that specific recording. For questions no single video answers, a library synthesizer works from distilled notes, drills into evidence and cites video plus timestamp for each finding. When it says the library does not clearly answer, take that at face value instead of stitching a guess from weak hits.

The split is simple: "which video showed X" is search, while "what did we decide about X" is library ask. The skill needs Bash and Read, so the watch-skill commands have to be available on your system. A playlist or a folder of recordings goes in with a single batch call.

What it helps with

  • find which video showed a specific error
  • recall what a meeting decided about a topic
  • search a phrase across every watched recording
  • answer a question that spans several videos
  • check whether a video was already analyzed

Example prompts for the agent

  • have we watched anything about this error?
  • which video had that slide, and at what timestamp?
  • what did we decide about the pricing page across those meetings?

How to install video-memory

1Skills CLI (Vercel)

npx skills add oxbshw/watch-skill --skill video-memory -a claude-code

2Manually: put the files in place

~/.claude/skills/video-memory/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/video-memory/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about video-memory

How do I install video-memory in Claude?

Run npx skills add oxbshw/watch-skill --skill video-memory -a claude-code. The skill then sits at ~/.claude/skills/video-memory/SKILL.md.

What does video-memory do?

It searches a persistent index of every analyzed video and answers questions from it: frames, OCR, transcripts and past answers all live in one library under ~/.watch-skill/.

How is search different from library ask?

search locates a video and timestamp for a phrase, while library ask answers a question no single video covers and cites video plus timestamp.

Source text

SKILL.md · MIT
---
name: video-memory
version: "1.4.3"
description: The user asks about videos watched in the past or across sessions — "have we watched anything about X", "which video showed that error", "what did that meeting decide", "search my videos", or a question that spans several videos. Use this to search and answer from the persistent cross-video index instead of saying you don't remember.
license: MIT
user-invocable: true
allowed-tools: Bash, Read
---

# Video memory

Every video ever watched on this machine is still here — frames, OCR,
transcripts, and past answers, in one persistent index under
`~/.watch-skill/`. You never have to say "I don't remember that video."
Sessions don't expire it; watching more videos makes it more useful.

## What's in the library?

```bash
watch-skill list
```

Every indexed video: id, title, duration, when it was analyzed. Check
here before watching anything — it may already be in memory from a
previous session.

## Find something across all of it

```bash
watch-skill search "<phrase or keywords>"
```

Hybrid keyword + semantic search across every video, any language
(Arabic folding, CJK and Thai segmentation are handled). Hits come back
with video + timestamp; follow up on a hit with:

```bash
watch-skill ask <video_id> "<the user's actual question>"
watch-skill ask <video_id> "what is on screen at <timestamp>?"
```

## Questions that span videos

"What did we decide about the pricing page across those meetings?" —
questions no single video answers go to the library synthesizer:

```bash
watch-skill library ask "<the question>"
```

It answers from distilled notes across every indexed video, drills into
real evidence, and cites video + timestamp for each finding. When it
says the library does not clearly answer, trust that — do not stitch a
guess together from weak search hits. `watch-skill library overview`
shows what the library knows (videos, note counts, entities recurring
across videos).

For "which video showed X" (locate, not synthesize), plain `search` is
the right tool; then `ask` the hit video.

## Batches build memory fast

A playlist or a folder of recordings goes in as one call —
`watch-skill batch "<playlist-or-folder>"` — and the whole set becomes
searchable memory (see `watching-videos`).

© oxbshw · MIT · oxbshw/watch-skill

Similar skills

All skills for Agent setup: memory, context, orchestration →

From the same set10

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor