The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown in the SKILL.md file. It does not include any executable scripts, binary files, or configuration that triggers network requests or file system modifications.
[INDIRECT_PROMPT_INJECTION]: While the skill manages decision-making based on external data (like PRDs or blueprints), it establishes clear boundaries and classification rules. It explicitly requires that security vulnerabilities and decisions contradicting user intent (User Challenges) are never automated and must be surfaced to the user. This design serves as a logic-level mitigation against potential manipulation of the decision flow.
[SAFE]: No obfuscation, hardcoded credentials, or persistence mechanisms were detected. The metadata matches the stated purpose of the skill.

auto-decision-framework