The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run to call external CLI tools such as acli, gog, and git. These interactions are securely implemented using list-based argument passing, which prevents shell injection vulnerabilities by ensuring that inputs (like Jira issue keys or spreadsheet IDs) are treated as data rather than executable code.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources including Jira issue descriptions and Google Sheet cells. This presents a potential surface for indirect prompt injection where malicious instructions could be embedded in the ingested content. The skill mitigates this through its architecture and instructions.
Ingestion points: Jira issue fields (key, summary, description) retrieved via scripts/_jira.py and Google Sheet content retrieved via scripts/release.py.
Boundary markers: The skill instructions mandate the use of Markdown tables and structured JSON for output, which helps enforce separation between data and instructions.
Capability inventory: The skill is strictly read-only and lacks capabilities to write to Jira or external services, limiting the impact of any potential injection.
Sanitization: The skill performs format validation, such as date parsing, on ingested data before processing.
[SAFE]: The skill is developed by redhat-developer and targets official Red Hat infrastructure (redhat.atlassian.net). The behavior is entirely consistent with its declared purpose of release status tracking.

rhdh-release-status