The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Python scripts (scripts/analyze-pr.py and scripts/triage-prs.py) that automate repository management tasks by executing GitHub CLI (gh) and git commands. Technical review of these scripts shows they use the subprocess.run method with arguments passed as a list and without the shell=True parameter, which is a secure practice that prevents common command injection vulnerabilities. The script inputs are further validated using the argparse module.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it ingests untrusted data from GitHub pull requests (such as titles, bodies, and labels) and possesses write capabilities like commenting on PRs and pushing code. This risk is effectively mitigated by the skill's instructions, which mandate the use of an external mutation-gate for every write operation, ensuring human review and approval before any external changes are made. The data processing is structured via Python scripts rather than relying on unstructured LLM scraping.
[EXTERNAL_DOWNLOADS]: The workflows include instructions to use curl for fetching backstage.json configuration files from raw.githubusercontent.com. These operations are used to verify version compatibility between upstream plugins and the local repository. The downloads target official GitHub infrastructure and known repositories, qualifying as safe references to well-known services under the established security guidelines.

rhdh-overlay