The Agent Skills Directory

[PROMPT_INJECTION] (LOW): Indirect Prompt Injection Surface. The skill ingests untrusted data from an external API which could theoretically contain malicious instructions.
Ingestion points: Market data (tickers, orderbooks) and account information (positions, transaction logs) retrieved from Bybit V5 API endpoints.
Boundary markers: Absent. The skill provides data to the agent in cleartext or formatted tables without explicit delimiters or instructions to ignore embedded commands.
Capability inventory: Significant trade execution capabilities, including order creation (/v5/order/create), order modification, and leverage adjustment.
Sanitization: API responses are parsed via Python for display, but no string-level sanitization is performed to neutralize potential prompt injection attacks contained within API fields like symbol names or error messages.
[COMMAND_EXECUTION] (SAFE): The skill utilizes curl and openssl for its core functionality of interacting with the exchange API. These operations are restricted to authorized Bybit domains and are necessary for the skill's stated purpose.
[DATA_EXFILTRATION] (SAFE): API credentials are used exclusively to generate signatures and are transmitted only to the official Bybit API endpoints via standard headers. No evidence of exfiltration to third-party domains was detected.

bybit-trader