content-writer

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Most links are benign documentation, images and well-known services, but the presence of multiple GitHub repos from small/unknown accounts plus a direct npm tarball (registry.npmjs.org/.../claude-content-writer-2.0.1.tgz) and install scripts that clone/run those repos are a notable delivery vector for malware if not vetted—treat as suspicious until the code/authors are verified.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). High-risk supply-chain/backdoor patterns: postinstall auto-runs an install script that git-clones external repos (claude-seo, humanizer) into the user's home and executes remote install scripts via execSync/bash without integrity checks or explicit user consent, and it also deletes/overwrites skill directories — this enables arbitrary remote code execution and a supply‑chain backdoor capable of exfiltration or system compromise.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The package's postinstall/install script auto-clones and installs external skill repositories (it runs git clone https://github.com/AgriciDaniel/claude-seo.git and git clone https://github.com/blader/humanizer.git and may run their install scripts), which fetches and executes remote code during setup and provides required runtime skills (claude-seo, humanizer) that the skill invokes in /writer:verify, so these URLs are used to pull remote code that directly affects runtime behavior.