The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill specifies a requirement to install the 'fallow' package globally using npm install -g fallow. Although npmjs.com is a well-known service, the package is installed without a specific version pin, which can introduce risks if a malicious version of the package is released to the registry.
[COMMAND_EXECUTION]: The skill uses shell commands to interact with the environment, including npm install, fallow audit, and git diff. While these tools are necessary for the skill's stated purpose of code review, the ability to execute commands and install software represents a significant capability tier.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process all files in a repository to identify bugs. This exposes the agent to indirect prompt injection if a project file contains malicious instructions designed to subvert the agent's logic during the review or 'auto-fix' phases.
Ingestion points: Reads all files in the current repository to apply its review checklist and run static analysis.
Boundary markers: The instructions do not define clear delimiters or 'ignore' instructions to prevent the agent from following directives embedded in the analyzed code.
Capability inventory: The skill has the ability to run shell commands (fallow, npm, git) and modify files on the local system.
Sanitization: There is no evidence of content sanitization or validation of the files before they are processed by the agent.

find-mismatch