Save Image

Download images, GIFs, and media from any URL using proper headers, Referer spoofing, and scraping to handle CDNs and social media sources correctly.

This skill appears to do what it says (download images with proper headers and simple scraping), but there are some gaps you should consider before installing: - Missing dependency declarations: SKILL.md expects external tools (gifgrep, yt-dlp, jq, grep, curl, file). Confirm those binaries are present on the host or ask the author to declare them and provide an install script. gifgrep and yt-dlp in particular can perform broad downloads and may have their own credential requirements. - Undeclared API key: references/tools.md mentions GIPHY_API_KEY for Giphy usage but the skill metadata doesn't list any required env vars. If you plan to use Giphy via gifgrep, obtain and provide that API key only when necessary and verify how the skill stores/uses it. - No hidden network endpoints were found in the included files, and the provided shell script only performs HTTP GETs and local file checks. Still, be mindful that scraping can retrieve login-gated or copyrighted content — avoid using private credentials or downloading material you do not have rights to. - Operational limits: the instructions note Cloudflare and login-gated content won't work; browser automation or authenticated sessions are required for those cases. Recommendations: ask the skill author to (1) list required binaries and env vars in the metadata, (2) provide installation instructions or package references for gifgrep/yt-dlp if the skill relies on them, and (3) confirm how any API keys are used and whether they are stored. Test the script on innocuous public images first.

Type: OpenClaw Skill Name: save-image Version: 1.0.0 The 'save-image' skill is designed to help an AI agent download media files correctly by handling common anti-bot measures like User-Agent and Referer headers. It includes a helper script (scripts/fetch-image.sh) and documentation (SKILL.md, references/tools.md) that outline a decision tree for using curl, yt-dlp, or gifgrep based on the source. The logic is transparent, follows best practices for shell scripting (e.g., set -euo pipefail), and contains no evidence of data exfiltration, persistence, or malicious execution.