Screenshots
PassAudited by ClawScan on May 1, 2026.
Overview
This instruction-only screenshot helper is coherent and purpose-aligned, with the main things to notice being local project files, persistent preferences/learnings, and vision-model review of screenshots.
This skill appears safe to use for its stated purpose. Before installing or invoking it, be comfortable with it creating a ~/screenshots workspace, retaining preferences/learnings there, and using vision review on screenshot images. Clear memory/learnings between sensitive clients or projects, keep raw captures free of real personal or secret data, and verify any marketing claims or social proof before using outputs publicly.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may create and update files in your home directory for screenshot projects, but the artifacts scope this to the ~/screenshots workspace and do not show destructive commands.
The workflow includes local filesystem mutations and a symlink to organize generated screenshot versions.
Create Project Folder: `mkdir -p ~/screenshots/{app-slug}/raw` ... Approval: `Symlink latest to current version`.Use a safe app slug, review paths before running shell steps, and keep the ~/screenshots workspace limited to intended screenshot assets.
Old preferences, client-specific notes, or inaccurate learned patterns could carry into later projects and remain after skill updates.
The skill stores persistent cross-project learnings and per-user preferences, then reuses them to influence future screenshot work.
Update `~/screenshots/learnings.md` ... `## Per-User Notes` ... `Before New Projects` check `learnings.md` for same app category and this user's preferences.
Review or clear ~/screenshots/memory.md and ~/screenshots/learnings.md between clients or projects, and avoid storing secrets or sensitive business details there.
If raw screenshots contain unreleased UI, personal data, or secrets, that content may be included in model-based visual review.
The workflow directs screenshot images to a vision-capable model for quality review before user presentation.
Use vision model to verify EVERY screenshot set ... If ANY check fails → fix before presenting to user.
Sanitize screenshots before use, follow the skill's own no-sensitive-data checklist, and confirm you are comfortable with model review of the images.