The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from GitHub issue titles, bodies, and comments. This creates an indirect prompt injection surface where a malicious issue could attempt to influence the agent's behavior during triage.
Ingestion points: SKILL.md (via list_issues, search_issues, issue_read tools and gh pr list output)
Boundary markers: Absent. The instructions do not specify the use of delimiters or warnings to ignore instructions within the retrieved data.
Capability inventory: Access to GitHub MCP tools for listing, searching, and reading repository data.
Sanitization: Absent. The skill suggests labels based on raw content.
[COMMAND_EXECUTION]: The skill uses the gh (GitHub CLI) tool within a PowerShell pipeline to filter pull requests.
Evidence: The workflow involves executing gh pr list --limit 100 --json ... and processing the JSON output locally.
Context: This is a standard administrative workflow for GitHub maintainers and relies on the user's existing local authentication.

github-issue-triage