The Agent Skills Directory

[General] (SAFE): The skill consists of structured markdown templates for coordinating sub-agents. It does not contain executable code itself, but rather orchestrates development tasks using clear role definitions and gatekeeping.\n- [Indirect Prompt Injection] (SAFE): While the skill ingests task descriptions from external plan files (Ingestion Point: feature-plan.md), it employs clear boundary markers (e.g., ## Task Description, ## Context) in the sub-agent prompts to isolate instructions from data. The process includes a mandatory verification step where reviewers are explicitly instructed to inspect actual code rather than trusting status reports (Sanitization: Independent code inspection).\n- [Dynamic Execution] (SAFE): The skill facilitates the creation and testing of code by sub-agents. This behavior is consistent with the primary purpose of a software development tool and is managed through a multi-stage review pipeline (Spec Review -> Code Quality Review) that acts as a quality and safety gate.

delegated-development