# Shot

> Captures and summarizes screenshots from the local desktop or a named window using the bundled capture script. Use when the user asks for a screenshot, screen capture, active-window capture, visual inspection of the current screen, or named-window screenshot. Not for browser-page screenshots when browser MCP tools are already controlling a page.

- Skill: `ryanthedev/shot` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add ryanthedev/shot`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ryanthedev/shot/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: ryanthedev (https://skillmd.com/u/ryanthedev)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ryanthedev/shot

---


# shot

Capture a local screenshot, then summarize the visual result without loading large image data unnecessarily.

## Workflow

1. Choose mode: `full`, `active`, or `window`.
2. From the plugin root, run `python3 skills/shot/scripts/capture.py --mode <mode>`. For a named window, add `--name '<window name>'`.
3. Inspect the returned thumbnail path when the user asks for visual interpretation.
4. Return the relevant summary and the saved screenshot path.

If the host has a subagent tool, pass the screenshot path to a fresh reviewer for visual summary. Otherwise use the host's image-inspection tool directly.

## Dependencies

The capture script requires Python packages `mss` and `Pillow` and may require OS screen-recording permissions.