The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (MEDIUM): The skill requires external CLI tools tokei and difft to be installed via brew or cargo. These are third-party binaries whose integrity and updates are managed outside the skill's environment.
[COMMAND_EXECUTION] (LOW): Utilizes the Bash tool to run diagnostic commands. Although the provided examples are restricted to tokei, difft, git, and jq, the capability to execute shell commands poses a latent risk if the agent is manipulated.
[PROMPT_INJECTION] (LOW): Vulnerable to indirect prompt injection. The skill ingests untrusted data from local files through analysis tools. Evidence Chain: 1. Ingestion points: Source code files processed by tokei/difft. 2. Boundary markers: Absent. 3. Capability inventory: Bash/subprocess execution. 4. Sanitization: Absent. Malicious instructions embedded in comments or code within an analyzed repository could be interpreted by the agent as high-priority commands when viewing the tool output.

code-stats