The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied problem statements for analysis across its reframing pipeline (ingestion point: SKILL.md). While this provides a surface for indirect prompt injection, the skill implements several quality gates, including the ASN (Actionability, Specificity, Novelty) test and Blind Spot Reports, to ground and validate the outputs. This data ingestion is integral to the primary purpose of the skill.
[COMMAND_EXECUTION]: The multi and ideate recipes describe the coordination of specialized subagents and CLI tools such as Codex and Antigravity. These capabilities are used for parallel brainstorming and are explicitly described as authorized components within the system's operational environment, showing no signs of unauthorized or malicious intent.

flux