rhdh-release-announce

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses subprocess.run to execute local CLI tools like acli (Atlassian CLI), gog (Google Workspace CLI), and git. These executions use argument lists rather than shell strings, which is a secure implementation that prevents shell injection attacks. The tools being executed are standard for the skill's purpose in a development environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources such as Jira issue descriptions and Google Sheets rows to populate Slack message templates.
  • Ingestion points: Data is read via _jira.py (Jira issue fields) and release.py (Google Sheets values).
  • Boundary markers: The skill does not explicitly use boundary markers for the data, but it employs /prose-editing for review.
  • Capability inventory: The skill can execute command-line tools (acli, gog, git) and write files.
  • Sanitization: Basic regex-based normalization is performed on versions and dates. The risk is mitigated by the 'human-in-the-loop' design where the agent only drafts a message for the user to copy, and explicitly instructs that it must not post to Slack automatically.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 12:09 PM
Security Audit — agent-trust-hub — rhdh-release-announce