The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by processing untrusted data from external sources (Perplexity queries) to generate configuration recommendations.
Ingestion points: External data enters the context via mcp__perplexity-ask__perplexity_ask in Phase 3 of the SKILL.md file.
Boundary markers: There are no explicit delimiters or instructions to treat the external tool output as untrusted content.
Capability inventory: The skill has extensive project access, including Read, Write, and Edit permissions for configuration files, as well as TaskCreate and TaskUpdate for workflow management.
Sanitization: There is no evidence of sanitization or escaping of the external content before it is processed for gap analysis.
Mitigation: The risk is mitigated by a mandatory human-in-the-loop step using AskUserQuestion before changes are applied, and a provided --dry-run argument that suppresses file modifications.
[EXTERNAL_DOWNLOADS]: Fetches the latest Claude Code features and changes from the official Anthropics GitHub repository's changelog.

evolving-config