The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's installation routine attempts to install a Python package named graphifyy. This name deviates from the legitimate graphify library by adding a trailing 'y' and does not match the approved vendor resource naming patterns for the author vanducng, indicating a risk of typosquatting.
[REMOTE_CODE_EXECUTION]: By downloading and installing the suspected typosquat package graphifyy, the skill enables the execution of remote code on the host system during the installation and import process.
[PERSISTENCE]: The skill provides automated installation of a post-commit git hook and a project-specific CLAUDE.md section. These mechanisms ensure that the skill's code is executed automatically during developer activities and remains active across agent sessions.
[INDIRECT_PROMPT_INJECTION]: The core functionality involves analyzing external data—such as code, documents, PDFs, images, and video transcripts—using LLM-powered subagents. This creates an attack surface where malicious instructions hidden in these files could compromise the agent's behavior.
[DYNAMIC_EXECUTION]: The skill generates and runs Python scripts at runtime via python -c blocks to manage extraction, merging, and querying. This dynamic construction and execution of code increases the risk of injection vulnerabilities.
[COMMAND_EXECUTION]: The skill executes system commands including git clone, yt-dlp, and various file system operations based on user-supplied paths and URLs, which could be exploited for command injection if inputs are not properly sanitized.

graphify