skills/whamp/skills/sideshow/Gen Agent Trust Hub

sideshow

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses npx -y sideshow to fetch and execute the Sideshow CLI tool from the npm registry. This is a standard method for utilizing Node.js tools and relies on a well-known service for package distribution.
  • [COMMAND_EXECUTION]: The skill interacts with the environment through shell commands executed via bash -lc. This allows it to publish content, manage sessions, and retrieve feedback as part of its core functionality.
  • [DATA_EXFILTRATION]: The skill facilitates the transfer of project data—such as code snippets, logs, and diagrams—to a user-configured SIDESHOW_URL. The instructions provide specific guidance on securely managing the SIDESHOW_TOKEN within the environment to prevent accidental exposure.
  • [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection via user feedback ingested through sideshow wait and sideshow watch in the primary skill loop. It mitigates this risk by instructing the agent to explicitly distinguish user comments from other ingested data and to treat only those comments as instructions, while noting the lack of automated sanitization at the transport level.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 01:23 PM
Security Audit — agent-trust-hub — sideshow