The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to execute git and gh (GitHub CLI) commands. These are used to manage the local environment, fetch PR data, and manipulate git worktrees for isolation.
[EXTERNAL_DOWNLOADS]: The skill downloads PR metadata and source code from the official nrwl/nx GitHub repository. This is an essential function for a code review tool and targets the vendor's own infrastructure.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection as it processes untrusted data from PR descriptions and comments to drive its review logic and subagent prompts.
Ingestion points: PR metadata (titles, bodies), PR comments, and linked issue timelines are retrieved via the gh tool and processed in SKILL.md (Step 4, Step 4.5).
Boundary markers: While the skill structures context using Markdown headers, it does not employ robust delimiters or specific instructions to subagents to ignore potentially malicious commands embedded in the PR data.
Capability inventory: The agent possesses broad capabilities including shell command execution, file system modification in the home directory, and the ability to invoke other skills and subagents.
Sanitization: There is no evidence of sanitization or escaping of the external PR content before it is interpolated into prompts or context files.

review-pr