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 commands/mnvsk97/eyeroll/watchgit clone --depth 1 https://github.com/mnvsk97/eyerollWrote 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/commands/mnvsk97/eyeroll/watch)<a href="https://agentmods.dev/commands/mnvsk97/eyeroll/watch"><img src="https://agentmods.dev/badge/commands/mnvsk97/eyeroll/watch.svg" alt="Measured on agentmods" 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 | $0.00000 | $0.01298 |
| Opus 5 | $0.00000 | $0.00649 |
| Sonnet 5 | $0.00000 | $0.00260 |
| Haiku 4.5 | $0.00000 | $0.00130 |
Grade A, and why
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 5d 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 — 132 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/eyeroll:watch
Analyze a video, screen recording, or screenshot and produce a structured report.
Usage
/eyeroll:watch <url-or-path> [--context "..."]
What it does
- Downloads the video (if URL) or reads the local file
- Extracts key frames (1 per 2 seconds, deduplicated, contrast-enhanced)
- Analyzes each frame with the configured vision model
- Transcribes audio (if backend supports it and audio is present)
- Checks cache for previous analysis of the same source
- Synthesizes a structured report using codebase context
The agent then presents a concise summary rather than dumping the raw report.
Arguments
| Argument | Description |
|---|---|
<url-or-path> |
Video URL (YouTube, Loom, etc.) or local file path |
--context "..." |
Additional context about the video (what it shows, what you want done) |
--no-context |
Skip auto-discovery of codebase context (CLAUDE.md, AGENTS.md, etc.) |
--no-cost |
Suppress cost estimate output |
--min-audio-confidence |
Minimum confidence for Whisper audio segments (0.0-1.0, default 0.4) |
--scene-threshold |
Pixel-diff threshold for scene-change frame extraction (default 30.0, 0=fixed interval) |
If you mention context in conversation without using the --context flag, the agent will pick it up and pass it along.
Examples
/eyeroll:watch https://loom.com/share/abc123
/eyeroll:watch ./bug-recording.mp4
/eyeroll:watch ./screenshot.png --context "this error shows up on the settings page"
/eyeroll:watch https://youtube.com/watch?v=xyz --context "implement this feature"
Report structure
The report first classifies the content type, then adapts its analysis sections accordingly.
Content type detection
eyeroll determines what kind of content the video shows:
| Content Type | Example | Analysis Focus |
|---|---|---|
| Bug report | Error on screen, broken behavior | Expected vs actual, fix directions, search patterns |
| Feature request | Mockup or desired behavior | What's requested, implementation notes, acceptance criteria |
| Question / answer request | User asks how something works | Direct answer, evidence, follow-up questions |
| Documentation lookup/update | Docs are being read or requested | Relevant docs, missing docs, suggested update |
| Tutorial | Step-by-step how-to | Process, tools used, skill automation potential |
| Code review | PR diff walkthrough | Files reviewed, concerns noted |
| Feature demo | Working feature walkthrough | Capabilities shown, codebase relevance |
| General notes | Meeting, brainstorm | Key takeaways, action items |
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.
- 5d ago First seen · 132 lines · 0 tokens per session scan A 7551de5de617
watch is a command published in the GitHub repository mnvsk97/eyeroll (17 stars, last pushed 3mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,298 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 commands, from other repositories
frontend
Injects context of all relevant cli files.
cleanup-back-to-main
Go back to main and clean up the branch.
context
项目上下文管理：初始化 .context 目录、记录决策日志、压缩归档、查看历史.
spec-research
需求 → 约束集（并行探索 + OPSX 提案）.
spec-design
Create comprehensive technical design for a specification.
edison.list
This obsolete slash command is disabled for this repository.