The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run inside scripts/collect.py to execute the gh command-line tool. This is the intended mechanism for gathering activity data. The script passes arguments as a list and does not invoke a shell, providing protection against basic command injection.
[EXTERNAL_DOWNLOADS]: The generated HTML output references Tailwind CSS and Chart.js from well-known content delivery networks (cdn.tailwindcss.com and cdn.jsdelivr.net). These resources are used for rendering and styling the interactive report.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted content from GitHub search results (e.g., issue titles, PR descriptions, and comments) to generate a narrative summary and assign work categories.
Ingestion points: Untrusted data enters the local environment through the gh CLI in scripts/collect.py and is saved to week.json.
Boundary markers: The skill does not implement explicit boundary markers or instructions to disregard potential injection markers within the GitHub content it processes.
Capability inventory: The skill maintains capabilities for file system writes (to the scratchpad and temp directories) and command execution via the gh CLI.
Sanitization: The report generation in template.html uses safe DOM methods such as textContent and append to insert external data, which effectively mitigates the risk of Cross-Site Scripting (XSS) in the generated HTML.

gh-weekly-report