The Agent Skills Directory

[SAFE]: The skill provides high-quality security documentation and best practices for smart contract developers. It covers critical areas such as access control, capability management, and on-chain randomness security without any malicious intent.
[NO_CODE]: The skill consists entirely of Markdown reference files and JSON evaluation prompts. No Python, JavaScript, or shell scripts are included, eliminating the risk of direct code execution or local environment compromise.
[EXTERNAL_DOWNLOADS]: The skill references an external MCP (Model Context Protocol) tool at https://sui.mcp.kapa.ai for documentation queries. This service is a standard integration for Sui technical documentation and is considered a safe, vendor-aligned resource.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and audit Move code provided by users. While processing untrusted code can theoretically expose the agent to indirect prompt injection, the skill's instructions are highly technical and provide specific auditing rules (e.g., verifying package IDs on-chain) that act as effective guardrails for the agent's behavior.

move-security