analyzing-market-sentiment

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEPROMPT_INJECTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill has a susceptibility to indirect prompt injection because it retrieves and processes article titles and summaries from external sources.
  • Ingestion points: Untrusted article content is fetched from CoinTelegraph, CoinDesk, and Decrypt RSS feeds in scripts/news_sentiment.py.
  • Boundary markers: The skill does not implement boundary markers or instructions to isolate processed text from agent instructions.
  • Capability inventory: The skill allows the use of Read and Bash(crypto:sentiment-*) tools, which could be targeted by instructions embedded in external news data.
  • Sanitization: Although basic HTML stripping is performed in _parse_rss_xml, there is no validation to detect or neutralize natural language injection attacks.
  • [REMOTE_CODE_EXECUTION]: The skill utilizes dynamic module loading, a pattern that warrants caution.
  • In scripts/news_sentiment.py, the skill dynamically modifies the Python execution path to import the NewsAggregator class from a sibling skill directory (crypto-news-aggregator), which constitutes dynamic loading from a computed path.
  • [EXTERNAL_DOWNLOADS]: The skill connects to several well-known financial and technology services to perform its analysis.
  • It retrieves the Fear & Greed Index from api.alternative.me.
  • It fetches market momentum data, including price and volume, from the CoinGecko API (api.coingecko.com).
  • It aggregates news articles from official RSS feeds provided by established crypto media outlets.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 06:58 AM