The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the GitHub CLI (gh) through localized Node.js scripts (github-board-report.mjs and setup-github-board.mjs). These scripts use execFileSync with argument arrays, which prevents shell-based command injection by ensuring arguments are passed directly to the subprocess without shell interpretation.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, specifically issue titles, descriptions, and metadata from GitHub and Jira. While this presents a surface for indirect prompt injection, the skill includes explicit instructions for the agent to treat all retrieved text as untrusted data and never as instructions. Furthermore, all write operations require explicit human approval, significantly mitigating the risk of automated exploitation.
[EXTERNAL_DOWNLOADS]: The skill interacts with well-known and trusted services, including GitHub and Atlassian (Jira) APIs. These interactions are fundamental to the skill's purpose and are conducted through official tools and authenticated connections, adhering to secure development practices.
[SAFE]: The implementation demonstrates strong security posture through several design choices: it defaults to read-only/dry-run modes, requires per-category approval for batches of changes, and strictly separates board-local configuration from broader organization-wide settings.

board