The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to run commands such as git status, git diff, and git log. These operations are used to manage Git state and analyze code history, which are powerful actions that could be abused if not handled with user confirmation (which the skill generally includes).
[PROMPT_INJECTION]: The skill exhibits vulnerability to indirect prompt injection as it reads and analyzes external data (Swift source code). Attacker-controlled code or comments could attempt to influence the agent's assessment logic or the output of the 'Issue Rating Table'.
[COMMAND_EXECUTION]: Data extracted from source files (specifically symbol names) is interpolated into shell strings, e.g., git log -p -S "symbolName". While Swift identifier syntax limits characters, the lack of explicit sanitization for shell metacharacters represents a vulnerability surface.
[PROMPT_INJECTION]: Mandatory Evidence Chain for Indirect Prompt Injection:
Ingestion points: Files matching **/*.swift are read by Grep, LSP, and Read tools.
Boundary markers: The skill does not implement delimiters or 'ignore' instructions for the content of the files it scans.
Capability inventory: The skill can execute shell commands via Bash, write reports via Write, and perform interactive actions via AskUserQuestion (including deleting files and committing to Git).
Sanitization: Extracted source content is used in reports and shell commands without explicit sanitization or escaping of the values.

dead-code-scanner