The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes the contents of repository-level files.
Ingestion points: The skill reads .github/pull_request_template.md (referenced in SKILL.md).
Boundary markers: No explicit boundary delimiters are defined for the template content.
Capability inventory: The skill uses git and gh CLI tools for local and remote repository operations.
Sanitization: The skill defines a robust defense against shell injection by mandating the use of HEREDOCs when passing user-provided content to git commit.
[DATA_EXPOSURE]: The skill explicitly instructs the agent to never commit secrets, supporting standard secure development practices.
[COMMAND_EXECUTION]: The skill utilizes git and gh for workflow automation. It includes specific patterns to handle multi-line strings safely in the shell, reducing the risk of unintended command execution from user inputs.

make-pr