The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The installation script clones the repository https://github.com/haibindev/repo-scan.git. Although the script uses a pinned commit hash (2742664ebcad1450c208eda0ae45d3c17fad5dd8) to ensure consistency, the source originates from a community account that is not listed as a trusted vendor, posing a potential supply chain risk.
[COMMAND_EXECUTION]: The skill contains a bash script that executes multiple system-level commands, including git clone, tar, mkdir, and mv. These commands are used to download external code and install it into the agent's local configuration directory ($HOME/.claude/skills/repo-scan).
[INDIRECT_PROMPT_INJECTION]: The intended purpose of the installed tool is to scan external source code repositories across various languages (C++, Java, Swift, etc.). This creates a vulnerability surface where a malicious repository being scanned could contain hidden instructions within code comments or metadata designed to influence the agent's classification or reporting behavior.
[DYNAMIC_EXECUTION]: The bootstrap process effectively adds new executable capabilities to the agent by installing external scripts and logic that are not contained within the primary SKILL.md file. The subsequent execution of the 'repo-scan' tool relies on these dynamically downloaded components.

repo-scan