The Agent Skills Directory

[COMMAND_EXECUTION]: The script utilizes the subprocess.run method to interact with git and the GitHub CLI (gh). It passes arguments as a list rather than a shell-interpreted string, which effectively mitigates command injection risks. These commands are essential for the skill's purpose of discovering and reading repository templates and metadata.
[EXTERNAL_DOWNLOADS]: The skill performs network operations by invoking the GitHub CLI to fetch issue bodies and repository contents. These operations target GitHub, a well-known and trusted service, and are performed using the user's authenticated CLI environment.
[INDIRECT_PROMPT_INJECTION]: This category identifies an inherent attack surface because the skill processes untrusted data from external sources (issue and pull request bodies). Although the script's logic is focused on structural Markdown parsing, an AI agent using the skill could be influenced by malicious instructions embedded within the text it analyzes. 1. Ingestion points: Untrusted data enters via the live_body function (retrieving GitHub content) and the run function (reading from local files or stdin). 2. Boundary markers: The script does not apply specific delimiters or instructions to ignore embedded content within the analyzed bodies. 3. Capability inventory: The skill has the capability to execute system commands (git, gh) and read local files. 4. Sanitization: The headings() function identifies ATX-style headings and checks for content but does not sanitize the text content for potential prompt injection patterns.

template-check