The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a methodology for the agent to analyze and refactor Rails codebases, including the 'Specification Test' and 'God Object Detection'. Processing untrusted source code creates a surface for indirect prompt injection if the code contains malicious instructions or adversarial data patterns.
Ingestion points: The agent is instructed to process project files within directories such as app/models/, app/controllers/, and app/services/ (specified in references/extraction-signals.md).
Boundary markers: The instructions do not specify any delimiters or safety prompts to ensure the agent ignores instructions embedded within the codebase being analyzed.
Capability inventory: The skill identifies capabilities such as shell command execution for complexity metrics and mentions the use of specialized tools like rails-refactorer for code modification.
Sanitization: There is no mention of sanitizing or validating code content before the agent evaluates it against the architectural rules.
[COMMAND_EXECUTION]: The skill provides specific shell command pipelines for the agent or developer to use when measuring code churn and complexity.
Evidence: references/extraction-signals.md contains complex shell commands using git log, sort, uniq, and wc (e.g., git log --since="6 months ago" --name-only --pretty=format: | sort | uniq -c | sort -rn | head -20). While intended for local diagnostics, these commands would be executed by the agent in the user's environment.

layered-rails