The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The script scripts/bootstrap_from_example.py downloads project templates by cloning the official luxonis/oak-examples repository from GitHub.
[COMMAND_EXECUTION]: The skill uses subprocess.run to execute the git clone command. The command is constructed using a list of arguments, which is a secure practice that avoids shell interpolation and injection risks.
[DATA_EXFILTRATION]: The skill creates a local cache directory at ~/.cache/luxonis/oak-examples to store the cloned repository content. No sensitive user data or system credentials are accessed or transmitted externally.
[PROMPT_INJECTION]: The skill processes and transforms AGENTS.md files from the cloned repository, which represents a surface for indirect prompt injection.
Ingestion points: Content is read from AGENTS.md files inside the cloned repository in scripts/bootstrap_from_example.py.
Boundary markers: No specific delimiters or boundary markers are used when interpolating this content into the generated project files.
Capability inventory: The script performs file system operations including directory creation (mkdir), file copying (shutil.copy2), and file writing (Path.write_text) in scripts/bootstrap_from_example.py.
Sanitization: While the script rewrites links and identifiers, it does not explicitly sanitize the source content for embedded instructions. However, the risk is mitigated by the use of the official Luxonis vendor repository as the data source.

bootstrap-example