The Agent Skills Directory

[SAFE]: No malicious logic, data exfiltration, or harmful code execution was detected across the skill's documentation and examples.
[PROMPT_INJECTION]: The skill ingests untrusted data from multiple external sources, which represents an inherent surface for indirect prompt injection. However, this is significantly mitigated by mandatory confirmation protocols.
Ingestion points: External data enters the agent's context through RSS feeds from trumpstruth.org, web scraping of farside.co.uk, and multiple cryptocurrency exchange API responses described in SKILL.md.
Boundary markers: Present and enforced. The skill contains clear instructions that no order, transfer, or funding action may be executed without an explicit 'CONFIRM' from the user.
Capability inventory: The skill has extensive capabilities to perform high-risk financial operations, including trading on eight exchanges and executing wallet transfers.
Sanitization: The instructions do not explicitly describe sanitization or filtering of external strings before display or processing.
[CREDENTIALS_UNSAFE]: The skill correctly instructs users to manage their secrets through a local .env file and does not contain any hardcoded credentials or instructions to share keys with the vendor.

blave-quant