# Loom Analyze

> Download a Loom video share URL and produce an audio transcript (whisper) plus optional keyframes for visual analysis. Triggers on any loom.com share URL the user provides, or on phrases "transcribe this loom", "analyze this loom", "watch this loom", or "/loom-analyze" followed by a URL.

- **Type:** Skill
- **Install:** `agentstack add skill-ao92265-claude-code-playbook-loom-analyze`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [ao92265](https://agentstack.voostack.com/s/ao92265)
- **Installs:** 0
- **Category:** [Content & Media](https://agentstack.voostack.com/c/content-and-media)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [ao92265](https://github.com/ao92265)
- **Source:** https://github.com/ao92265/claude-code-playbook/tree/main/skills/loom-analyze

## Install

```sh
agentstack add skill-ao92265-claude-code-playbook-loom-analyze
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Loom Analyze

Local pipeline (no API key, no third-party MCP) that turns a Loom share URL into:

1. A plain-text transcript via `whisper`.
2. Optional keyframe PNGs for visual analysis when the user asks about UI / screen content.

## When to invoke

- User pastes a `https://www.loom.com/share/...` URL.
- User says "transcribe", "analyze", or "watch this loom".
- `/loom-analyze ` is typed.

## How to run

The shell wrapper lives next to this file: `loom-analyze.sh`. Always call it through Bash.

```bash
${CLAUDE_PLUGIN_ROOT:-$HOME/.claude}/skills/loom-analyze/loom-analyze.sh  [--model M] [--frames N] [--keep]
```

Flags:

- `--model base|small|medium|large` — whisper accuracy/speed tradeoff. Default `base` (~140 MB download first time).
- `--frames N` — also extract one PNG every N seconds into `frames/`. Skip this for transcript-only (faster).
- `--keep` — keep MP4/MP3 after transcribing. Default deletes them.

Output goes to `~/Downloads/loom-transcripts//`. The transcript is printed to stdout, so the agent gets it in the tool result.

## Decision rules for the agent

- **Transcript-only is the fast default.** Don't request frames unless the user asks about visuals, UI, screen content, or a demo walkthrough.
- **Long videos (>10 min):** stick to `--model base`. `--model small` or larger only when the user specifically complains about transcript accuracy.
- **When frames are extracted:** read the PNGs from `~/Downloads/loom-transcripts//frames/` via the Read tool when the user asks what's on screen at a given time. Don't bulk-read every frame; pick the ones that match the question.
- **First-time setup:** if the script exits with "Missing dep", run `${CLAUDE_PLUGIN_ROOT:-$HOME/.claude}/skills/loom-analyze/setup.sh` once, then retry.

## Setup (per-machine, one-time)

```bash
bash ${CLAUDE_PLUGIN_ROOT:-$HOME/.claude}/skills/loom-analyze/setup.sh
```

Installs `yt-dlp`, `ffmpeg`, `openai-whisper` via Homebrew + pip. Idempotent.

For machine requirements, model disk sizes, troubleshooting, and limitations, see `references/usage.md`.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [ao92265](https://github.com/ao92265)
- **Source:** [ao92265/claude-code-playbook](https://github.com/ao92265/claude-code-playbook)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-ao92265-claude-code-playbook-loom-analyze
- Seller: https://agentstack.voostack.com/s/ao92265
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.