The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill instructs the agent to automatically switch to a code-execution skill when refactoring more than 10 files. It provides a specific Python template (from api.code_transform import rename_identifier; result = rename_identifier(...)) to be executed, which involves assembling a script using identifiers provided by the user.
[INDIRECT_PROMPT_INJECTION]: The skill processes and modifies existing source code files, creating a surface for indirect prompt injection from data (such as malicious comments or strings) within those files.
Ingestion points: The skill reads code occurrences and surrounding context using the Grep tool as defined in the native workflow (SKILL.md).
Boundary markers: No specific delimiters or instructions are provided to the agent to help it distinguish between legitimate code and potential instructions embedded within the code files being refactored.
Capability inventory: The skill possesses capabilities for file searching (Grep), file modification (Edit), and arbitrary code execution through the integrated code-execution skill.
Sanitization: The instructions do not define any sanitization or validation of the code content before it is processed or used to construct refactoring operations.
[COMMAND_EXECUTION]: The skill utilizes the Grep command-line utility to locate identifiers and patterns across the filesystem and to verify changes after they have been applied.

code-refactor