The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitHub pull requests (titles, summaries, and bodies) which are then rendered into an HTML dashboard. While the agent primarily opens the dashboard for the user, this ingestion represents a vulnerability surface where malicious instructions could be embedded in PR content.
Ingestion points: Data is fetched from the GitHub API using gh api in scripts/pr-dashboard-cli.mjs.
Boundary markers: The skill lacks explicit instructions or delimiters to prevent the agent from obeying potential instructions found within pull request descriptions.
Capability inventory: The skill executes shell commands via the GitHub CLI, writes HTML files to the temporary directory, and opens the system's default browser.
Sanitization: The script uses an escapeHtml utility to sanitize fields like repository names and PR titles, and utilizes GitHub's official Markdown API to render PR bodies, reducing the risk of XSS in the generated dashboard.
[COMMAND_EXECUTION]: The skill executes the GitHub CLI (gh) to fetch data and uses system commands (open, xdg-open, or start) to display the results. It correctly uses execFile and spawn without a shell to mitigate command injection risks from user-supplied date ranges or roles.

pr-dashboard