creator-authenticity-score
Fail
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: HIGHCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill's setup instructions for Path A involve Python code that executes shell commands using
subprocess.run. These commands are used to configure themcportertool with an access token retrieved from the vendor's API.\n- [EXTERNAL_DOWNLOADS]: The authentication flow described in the skill performs network requests tomcp.xpoz.aifor dynamic client registration and token exchange. These operations are necessary for setting up access to the vendor's Xpoz service.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external social media sources, which presents a surface for indirect prompt injection.\n - Ingestion points: External content is retrieved using the
getInstagramCommentsByPostIdtool, which fetches comment text from Instagram posts.\n - Boundary markers: The instructions do not include specific delimiters or warnings to the agent to ignore instructions that might be embedded within the retrieved comment text.\n
- Capability inventory: The skill environment, as described in the setup instructions, allows for network requests (
urllib.request.urlopen), local file system writes (~/.cache/xpoz-oauth/state.json), and the execution of shell commands (subprocess.run).\n - Sanitization: There is no evidence of sanitization, filtering, or validation of the external comment text before it is presented to the agent for analysis.
Recommendations
- HIGH: Downloads and executes remote code from: https://mcp.xpoz.ai/oauth/token - DO NOT USE without thorough review
Audit Metadata