Tapestry Skills: why it is not safe to use
Tapestry Skills passed our review at 82/100, with two cautions: web pages it reads could carry injected instructions, and it can auto-install system tooling.
What we found
Tapestry Skills reads user-supplied URLs (YouTube videos, articles, PDFs) and writes the extracted text to local files. In our scan it stored content locally only, carried no hardcoded secrets, made no hidden network calls, and did not touch credentials, environment variables, or browser data. It cleared those checks and scored 82/100, a pass.
Two items drew a caution rather than a failure. First, downloaded article text and video transcripts are fed back to the agent, so a malicious page could embed instructions the agent then reads. This is inherent to any read-a-URL tool and is no worse here. Second, the skill can auto-install tooling (yt-dlp, poppler, a readability extractor, optional Whisper) through brew, apt, npm, or pip, which modifies your local system. The installer script itself only copies files.
What to do instead
Point it at sources you trust, and review content it processes from unfamiliar pages. Install the required tools yourself beforehand so nothing is added without your knowledge, and run the skill under an account without elevated privileges.
Source: https://github.com/michalparkola/tapestry-skills-for-claude-code
We report what our security review found at the time we checked, with the goal of keeping people safe. Projects change; if a maintainer has since fixed this, we are glad to recheck it. Email hello@agentpod.com.