The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation includes an installation command for the Composio CLI: curl -fsSL https://composio.dev/install | bash. This fetches and executes a script from the vendor's domain to set up the necessary environment.
[COMMAND_EXECUTION]: The agent is instructed to perform various shell-based tasks including repository searches (rg), local code transformations using AST-based tools, and Git operations such as branching, committing, and pushing code.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection (Category 8) as it programmatically reads and modifies files from the user's codebase.
Ingestion points: Local source files identified through search patterns and passed to refactoring tools (e.g., jscodeshift).
Boundary markers: The instructions do not define specific delimiters or security warnings to prevent the agent from obeying instructions embedded in the code being refactored.
Capability inventory: The skill allows for file modification, execution of local build/test tools, and interaction with remote repository hosting and project management services.
Sanitization: There is no explicit logic described for sanitizing content retrieved from local files before processing or committing it.

codebase-migrate