The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses extensive Bash scripting to analyze file content and names. Many scripts, such as those in modules/quick-scan.md and modules/ai-generated-bloat.md, lack proper quoting around variables derived from user-controlled data (e.g., pkg, file, func). This could allow for command injection or execution errors if a codebase contains maliciously crafted filenames or strings.
[PROMPT_INJECTION]: The skill's architecture is vulnerable to indirect prompt injection because it ingests untrusted code and documentation.
Ingestion points: Code files (.py, .js, .ts), markdown documentation, and git history records are parsed across all modules (Tier 1-3).
Boundary markers: No specific delimiters or safety instructions are used to separate untrusted content from the agent's logic or the shell commands it constructs.
Capability inventory: The agent can execute shell commands, read/write files via the Bash and Read tools, and invoke external analysis utilities like vulture or knip.
Sanitization: There is no evidence of sanitization or strict validation of the content extracted from the files before it is passed to shell pipelines (e.g., in ai-generated-bloat.md where import strings are parsed and used in shell commands).

bloat-detector