The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted project data including code diffs, CI configs, and internal documentation without using delimiters or sanitization. This establishes a vulnerability surface where malicious code changes could attempt to override agent instructions. Ingestion points: Code files, git diffs, and configuration files. Boundary markers: Absent. Capability inventory: Shell command execution for tests and linter, file writing for evidence, and source code modification. Sanitization: None provided.
[COMMAND_EXECUTION]: The skill executes arbitrary project-specific commands for linting and testing as specified in the repository's configuration.
[DYNAMIC_EXECUTION]: The skill dynamically discovers and executes scripts and 'fix/check modes' at runtime based on the package manager and configuration files detected in the project path.

code-cleanup