The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it processes user-provided text.
Ingestion points: Untrusted data enters via the input.text field in SKILL.md.
Boundary markers: Absent; user input is not enclosed in delimiters to prevent instruction leakage.
Capability inventory: Performs network POST requests to Binance API through the skill-router configuration.
Sanitization: No input validation or sanitization is performed on the user content before submission.
[SAFE]: Implements secure credential management. The skill uses an agent-secrets.yaml configuration to define a slot for the Binance Square OpenAPI key, ensuring secrets are managed by the platform rather than hardcoded.
[SAFE]: Network activity is restricted to well-known domains. The skill communicates with official Binance endpoints (www.binance.com), which is consistent with its stated purpose.

square-post