skills/terminalskills/skills/val-town/Gen Agent Trust Hub

val-town

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill uses Val Town's standard library and user modules imported via URLs from the official CDN (https://esm.town). This is the platform's native method for managing code dependencies.
  • [COMMAND_EXECUTION]: Provides templates for serverless functions that perform network operations via fetch and database operations through Val Town's built-in SQLite integration.
  • [PROMPT_INJECTION]: The skill processes untrusted external data, which creates a potential surface for indirect prompt injection.
  • Ingestion points: The skill ingests data from HTTP request bodies (req.json(), req.text()), URL parameters (url.searchParams), and external API responses in SKILL.md.
  • Boundary markers: None explicitly defined in the provided code templates to differentiate between data and instructions.
  • Capability inventory: The functions demonstrate network access via fetch and persistent data storage via sqlite in SKILL.md.
  • Sanitization: The examples provided are generic templates and do not include specific input validation or sanitization logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 04:32 PM
Security Audit — agent-trust-hub — val-town