The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from external web sources (via exa and context7) and the local codebase, which presents a surface for indirect prompt injection. This risk is effectively mitigated by mandatory human-in-the-loop checkpoints in Phase 6 (Implementation Plan Approval) and Phase 8 (Code Review), ensuring the user validates all proposed actions and code before execution.
[EXTERNAL_DOWNLOADS]: The skill utilizes well-known research tools (exa and context7) to perform web searches and retrieve documentation. These operations are transparently part of the 'External Resource Research' phase and are intended to gather technical information for the development process.
[COMMAND_EXECUTION]: The workflow involves orchestrating specialized background agents for code exploration and depth-focused code review. These processes are internal to the platform environment and are managed through structured Task List tools, providing visibility into the agent's activities.

requirement-analysis