/init
Set up eyeroll for this project — pick a backend, install dependencies, and generate codebase context
$ npx -y skills add mnvsk97/eyeroll --agent claude-codeHow it fires
How this command gets triggered: by you, by Claude, or both.
- Fires itselfClaude auto-loads it when your prompt matches the work.
- You can call itInvoke it directly when you want it.
- Slash command
/init
Context preview
What this command does when you run it.
Set up eyeroll for this project — pick a backend, install dependencies, and generate codebase context
Command definition
init.mddescription: Set up eyeroll for this project — pick a backend, install dependencies, and generate codebase context
argument-hint: ''
allowed-tools: Bash, Read, Write, Glob, Grep, AskUserQuestion
Set up eyeroll for the current project.
Step 1: Check if eyeroll is installed
eyeroll --help
If not found, ask the user which backend they want:
Use `AskUserQuestion` with these options:
- `Gemini (Recommended)` — Google Gemini Flash API, best quality, requires API key or credentials.json
- `OpenAI` — GPT-4o, requires API key
- `Ollama` — Local models, private, no API key needed
Then install:
- Gemini: `pip install eyeroll[gemini]`
- OpenAI: `pip install eyeroll[openai]`
- Ollama: `pip install eyeroll`
Step 2: Run eyeroll init
eyeroll init
This is interactive — it asks for the backend choice and API key. Let the user interact with it directly.
Step 3: Generate codebase context
After setup, explore the codebase and write `.eyeroll/context.md`:
1. Read `CLAUDE.md`, `README.md`, `pyproject.toml` or `package.json` 2. Run `git log --oneline -10` and `git branch --show-current` 3. Run `ls` on root and key source directories 4. Skim 2-3 key source files
Write `.eyeroll/context.md` with this structure — keep it under 80 lines:
# Project: <name>
## Stack
<language, framework, key libraries>
## Structure
<directory layout>
## Key Files
- <file>: <purpose>
## Recent Activity
<last 5-10 commits, current branch>
## Notes
<anything relevant — conventions, known issues>
Also write `.eyeroll/context_meta.json` with the generation timestamp:
{"generated_at": "<current UTC ISO timestamp>"}This lets eyeroll detect when the context goes stale (newer commits than the context).
**Note:** If the project already has `CLAUDE.md`, `AGENTS.md`, or similar coding tool context files, eyeroll auto-discovers those at watch-time — `.eyeroll/context.md` is only needed as a fallback when no such files exist.
Step 4: Confirm
Tell the user:
- eyeroll is ready
- `.eyeroll/context.md` has been generated (with staleness tracking)
- Codebase context is auto-discovered from CLAUDE.md, AGENTS.md, etc. if present
- They can run `/eyeroll:watch <video>` to analyze a video
Read more
description: Set up eyeroll for this project — pick a backend, install dependencies, and generate codebase context argument-hint: '' allowed-tools: Bash, Read, Write, Glob, Grep, AskUserQuestion
Set up eyeroll for the current project.
Step 1: Check if eyeroll is installed
eyeroll --help
If not found, ask the user which backend they want:
Use `AskUserQuestion` with these options:
- `Gemini (Recommended)` — Google Gemini Flash API, best quality, requires API key or credentials.json
- `OpenAI` — GPT-4o, requires API key
- `Ollama` — Local models, private, no API key needed
Then install:
- Gemini: `pip install eyeroll[gemini]`
- OpenAI: `pip install eyeroll[openai]`
- Ollama: `pip install eyeroll`
Step 2: Run eyeroll init
eyeroll init
This is interactive — it asks for the backend choice and API key. Let the user interact with it directly.
Step 3: Generate codebase context
After setup, explore the codebase and write `.eyeroll/context.md`:
1. Read `CLAUDE.md`, `README.md`, `pyproject.toml` or `package.json` 2. Run `git log --oneline -10` and `git branch --show-current` 3. Run `ls` on root and key source directories 4. Skim 2-3 key source files
Write `.eyeroll/context.md` with this structure — keep it under 80 lines:
# Project: <name> ## Stack <language, framework, key libraries> ## Structure <directory layout> ## Key Files - <file>: <purpose> ## Recent Activity <last 5-10 commits, current branch> ## Notes <anything relevant — conventions, known issues>
Also write `.eyeroll/context_meta.json` with the generation timestamp:
{"generated_at": "<current UTC ISO timestamp>"}This lets eyeroll detect when the context goes stale (newer commits than the context).
**Note:** If the project already has `CLAUDE.md`, `AGENTS.md`, or similar coding tool context files, eyeroll auto-discovers those at watch-time — `.eyeroll/context.md` is only needed as a fallback when no such files exist.
Step 4: Confirm
Tell the user:
- eyeroll is ready
- `.eyeroll/context.md` has been generated (with staleness tracking)
- Codebase context is auto-discovered from CLAUDE.md, AGENTS.md, etc. if present
- They can run `/eyeroll:watch <video>` to analyze a video
AI eyes that roll through video footage — watch, understand, act. eyeroll is a Claude Code plugin that analyzes screen recordings, Loom videos, YouTube links, and screenshots, then helps coding agents fix bugs, build features, and create skills.
Other commands on eyeroll.
- /api-watch
Analyze a video URL or local file using the eyeroll hosted API.
Open command - /fix
Watch a bug video, diagnose the issue, fix the code, and raise a PR
Open command - /history
List past video analyses from the cache
Open command - /watch
Analyze a video, screen recording, or screenshot and produce a structured report
Open command

