The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to facilitate Git operations.
Employs git ls-remote with security-hardened flags (disabling terminal prompts and credential helpers) to probe repository visibility (SKILL.md).
Executes git clone to perform the final repository download following user approval (SKILL.md).
[EXTERNAL_DOWNLOADS]: The skill is designed to interact with external Git hosts to retrieve source code. This is its primary intended function, and it includes a security assessment step via the oss-security-check skill to mitigate risks associated with untrusted public code.
[PROMPT_INJECTION]: The skill processes user-supplied repository URLs and shorthands, which are interpolated into shell commands. This represents a potential surface for indirect prompt injection or command injection if input normalization is bypassed.
Ingestion points: Processes repo URLs and shorthand (e.g., org/repo) from user requests (SKILL.md).
Boundary markers: No explicit delimiters are used to wrap the <normalized-https-url> variable within the shell command block.
Capability inventory: Executes shell commands via the git binary (SKILL.md).
Sanitization: The workflow specifies normalizing inputs to HTTPS URLs, which serves as a basic validation step, though it does not explicitly detail shell escaping or strict regex validation of the host/path components.

git