news-sentiment-engine
Warn
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's setup instructions direct the user to clone an external repository from a community source (
https://github.com/tellmefrankie/news-engine). - [REMOTE_CODE_EXECUTION]: Following the download, the skill instructs the user to run the project using
pnpm dev. Although the instructions specifypnpm install --ignore-scriptsto prevent execution of arbitrary scripts during the installation phase, the final execution phase still relies on unverified external code. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from multiple external RSS feeds (e.g., TechCrunch, The Verge, Hacker News).
- Ingestion points: RSS feeds in SKILL.md.
- Boundary markers: The instructions lack explicit delimiters or specific guardrail prompts to prevent the agent from following malicious instructions potentially embedded in the news articles.
- Capability inventory: The skill utilizes Claude for sentiment analysis and ranking, which can be influenced by the content of the ingested data.
- Sanitization: There is no evidence of content filtering or sanitization before the data is processed by the agent.
Audit Metadata