The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the bash tool to execute development workflows, including running unit tests (bun test, npm test, pytest), performing type checks (tsc --noEmit), and checking for linting errors (eslint). These are standard practices for verifying refactoring steps.
[COMMAND_EXECUTION]: The instructions direct the agent to execute a local Python script scripts/ast_grep_helper.py to perform structural code searches and replacements. This represents a capability to execute local scripts to augment the agent's refactoring capabilities.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests and processes external codebase content that is then used to drive high-privilege operations such as file writing (edit, lsp_rename) and command execution (bash).
Ingestion points: User-provided refactoring-target arguments and codebase files discovered by explore subagents.
Boundary markers: The skill does not explicitly define delimiters to isolate ingested code content from the agent's instructions, though it emphasizes 'understanding intent' as a manual gate.
Capability inventory: The agent utilizes bash for testing and edit for file modifications across the project.
Sanitization: No explicit sanitization of codebase content is performed before the agent processes it for refactoring logic.

refactor