The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill's primary purpose is to ingest and analyze untrusted data (code diffs and file contents). This creates a surface for indirect prompt injection where an attacker could place malicious instructions in code comments or strings to manipulate the agent's review results.
Ingestion points: The skill uses git diff and the Read tool to ingest file contents into the agent context for review.
Boundary markers (Absent): The skill does not define specific delimiters or instructions to the agent to ignore embedded instructions within the code it reviews.
Capability inventory: The agent has access to mcp__github__* tools to post issue comments and manipulate PRs, and shell access for git and gh CLI commands.
Sanitization (Present): The skill uses a quoted heredoc pattern (<<'EOF') when constructing the shell command for gh issue comment. This is a strong security practice that prevents the content of the review artifact from being interpreted as code or subshells by the local shell environment.
Command Execution (SAFE): The skill uses standard local tools (git, grep, gh) to perform its tasks. The shell command templates provided in the instructions use best practices to avoid shell injection when handling dynamically generated text.

comprehensive-review