The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes shell commands such as git diff and git status to retrieve code changes and repository state for analysis.
[DATA_EXFILTRATION]: The skill reads local project files and code diffs, then uses the mcp-gitlab-usage tool to transmit summaries and verdicts to an external GitLab instance. While this is the intended purpose for remote code review, it establishes a data flow from the local environment to a remote service.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted data from code diffs and GitLab issues, which can contain hidden instructions.
Ingestion points: Data enters the context via git diff outputs and GitLab issue descriptions accessed through ISSUE_URL.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the diffs or issue content.
Capability inventory: The skill can write files to the local directory (report generation in docs/versions/) and perform network-mediated actions like posting comments to GitLab.
Sanitization: There is no explicit logic to sanitize or escape the content of the diffs or external issue descriptions before analysis.

ns-code-reviewer