The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches content from external, user-defined RSS feeds in signal_sniper.py using urllib.request.urlopen. It implements a validate_url function to mitigate Server-Side Request Forgery (SSRF) by blocking localhost and private IP ranges.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted article headlines and summaries to determine trading actions.
Ingestion points: Untrusted text enters the system through the fetch_rss function in signal_sniper.py which retrieves data from external URLs.
Boundary markers: The skill does not use specific delimiters or instructions to prevent the agent from obeying instructions potentially hidden within the RSS feed content.
Capability inventory: The skill possesses high-impact capabilities, including the ability to execute financial trades (execute_trade) and redeem wallet balances.
Sanitization: While the script uses keyword-based sentiment analysis (_count_keyword_hits), the AI agent is instructed in SKILL.md to analyze the article content directly, which creates an opening for adversarial text to influence the agent's logic and trade execution.

polymarket-signal-sniper