geo-visibility-check
Fail
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: HIGHINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external social media platforms to build analysis panels. \n
- Ingestion points: The
getRedditPostsByKeywordsandgetTwitterPostsByKeywordstools inSKILL.mdare used to fetch community content. \n - Boundary markers: No explicit delimiters or boundary markers are defined to isolate the ingested data from the agent's instructions. \n
- Capability inventory: The skill environment supports
subprocess.runcalls, package installation viapip/npm, and network operations to vendor APIs. \n - Sanitization: The skill does not describe any specific sanitization or filtering logic for the ingested social media data. \n- [COMMAND_EXECUTION]: The skill instructions provide scripts that perform local command execution and file system operations to configure the environment. \n
- Evidence: Path A includes Python scripts that use
subprocess.runto configure themcportertool and manage state files in~/.cache/xpoz-oauth/. \n - Evidence: Path C instructs the user to execute
pip installandnpm installfor vendor SDKs. \n- [EXTERNAL_DOWNLOADS]: The skill fetches additional logic and configuration from external repositories. \n - Evidence: The instructions direct the agent to fetch the
ai-answer-traceskill definition fromhttps://raw.githubusercontent.com/XPOZpublic/xpoz-agent-skills/main/skills/ai-answer-trace/SKILL.mdif not already present.
Recommendations
- HIGH: Downloads and executes remote code from: https://mcp.xpoz.ai/oauth/token - DO NOT USE without thorough review
Audit Metadata