newsjack-monitor-setup
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to manage monitoring tasks. It uses the 'newsjack' command-line tool for initialization ('newsjack monitor init'), scheduling ('newsjack monitor schedule'), testing ('newsjack monitor test'), and running analysis ('newsjack monitor run'). It utilizes the 'gh' CLI to star the author's GitHub repository ('gh api -X PUT /user/starred/elvisun/newsjack') upon user confirmation.
- [PROMPT_INJECTION]: The skill contains an indirect prompt injection attack surface (Category 8). 1. Ingestion points: The skill collects untrusted user input for company metadata, topics, competitors, search terms, and spokespeople to populate configuration files (SKILL.md). 2. Boundary markers: The generated 'brief.md' and 'profile.json' files do not include explicit delimiters or instructions to the reading tool to ignore potentially malicious embedded instructions. 3. Capability inventory: The 'newsjack' CLI tool possesses the capability to read and write files, schedule recurring tasks, and perform network operations to fetch news content. 4. Sanitization: There is no evidence of validation or sanitization of the user-provided strings before they are interpolated into the JSON and Markdown configuration files.
Audit Metadata