The Agent Skills Directory

Privilege Escalation (HIGH): The scripts/install-deps.sh file contains a command to execute sudo npm install -g readability-cli. Requiring administrative privileges for dependency installation is a high-risk practice in AI agent environments.
Indirect Prompt Injection (HIGH): This skill has a high-risk attack surface for indirect prompt injection.
Ingestion points: The skill fetches and processes content from arbitrary user-provided URLs.
Boundary markers: There are no explicit instructions or delimiters mentioned to prevent the agent from obeying instructions embedded within the extracted article content.
Capability inventory: The skill executes shell scripts (extract-article.sh), performs network requests, and writes files to the local filesystem.
Sanitization: Since the source for extract-article.sh is missing, sanitization of extracted content cannot be verified. Malicious instructions in a webpage could potentially trigger unintended agent actions after extraction.
Unverifiable Behavior (MEDIUM): The primary logic resides in scripts/extract-article.sh, but the content of this script is not provided in the skill package. This prevents a full audit of how URLs are handled, how subprocesses are spawned, and how file paths are sanitized.
External Downloads (LOW): The skill installs trafilatura (Python) and readability-cli (Node.js). These are standard tools, but their installation is performed via unversioned commands (pip install, npm install), which could lead to non-deterministic or supply-chain risks.

article-extractor