The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script (detect_lang.py) to analyze the repository structure and identify programming languages. It also provides shell commands for the user to manually install the pre-commit framework.
[EXTERNAL_DOWNLOADS]: The skill generates GitHub Actions workflow templates that utilize well-known official actions from the actions (GitHub) and arduino organizations to configure the environment and run linting tasks.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection as it reads and parses existing configuration files within the target repository. 1. Ingestion points: Processes repository contents and configuration files in SKILL.md (Steps 1, 3, and 4). 2. Boundary markers: None explicitly defined in the provided logic. 3. Capability inventory: Local file write and merge operations for configuration files. 4. Sanitization: The risk is mitigated by an interactive 'Controlled Workflow' that inventories all changes and requires explicit user approval before modifying any files.

codeguard-init