The Agent Skills Directory

[SAFE]: The skill implements robust credential management by recommending the use of environment variables and .env files. It enforces strict rules against the agent ever displaying or hardcoding real API keys or secrets in any output or code blocks.
[SAFE]: Implements a mandatory 'Structured Operation Confirmation' flow for all mainnet write operations. This requires the user to manually type 'CONFIRM' after reviewing a summary card and explicitly forbids the agent from accepting automated or self-generated confirmations.
[SAFE]: Features an auto-update and modular routing system that downloads instructional markdown files from the vendor's official GitHub repository (bybit-exchange/skills). This system includes robust security controls such as strict path validation to prevent traversal and SHA256 checksum verification to ensure content integrity.
[SAFE]: Proactively addresses indirect prompt injection by identifying untrusted data sources (such as P2P chat messages, order notes, and market annotations). It instructs the agent to wrap this data in code blocks, treat it as display-only, and never interpret it as instructions.
[SAFE]: Includes safety features like 'Large Trade Protection,' which provides additional warnings for orders that exceed defined balance or dollar thresholds.

bybit-trading