The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it is designed to analyze third-party code, which may contain malicious instructions embedded within code comments or string literals.
Ingestion points: Processes untrusted external inputs via the $TASK variable within the pre hook configuration in SKILL.md.
Boundary markers: Absent; the skill lacks specific boundary delimiters or system instructions instructing the agent to treat data inputs strictly as data and ignore any embedded command directives.
Capability inventory: Possesses capabilities to log data via hooks and interface with external tool bindings such as mcp__claude-flow__memory_usage and mcp__claude-flow__github_repo_analyze.
Sanitization: No input sanitization, filtering, or content escaping is implemented prior to processing the task data.

agent-reviewer