The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to suggest an external command to the user: npx skills add wondelai/skills/<slug> --global. This pattern involves downloading and executing code from an unverified third-party repository, which presents a supply chain risk.
[COMMAND_EXECUTION]: The skill utilizes shell commands such as git log to analyze project history (churn analysis) and explicitly suggests using the npx utility for skill management.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from the user's project codebase and git history. Maliciously crafted comments or code could attempt to influence the agent's behavior during the analysis or refactoring phases.
[PROMPT_INJECTION]: Indirect injection evidence chain:
Ingestion points: The skill reads local project files, the docs/ directory, and the output of git log (File: SKILL.md).
Boundary markers: While the skill suggests reading before writing, it lacks specific delimiters or system instructions to ignore embedded commands within the analyzed code.
Capability inventory: The skill has the ability to read and write to the filesystem and execute shell commands.
Sanitization: No mechanisms for sanitizing or escaping content from the analyzed codebase are defined.
[EXTERNAL_DOWNLOADS]: The skill references the wondelai/skills repository for fetching additional modular skills. This repository is not identified as a trusted vendor or well-known service.

improve-code-quality