convex-check-updates

Fail

Audited by Gen Agent Trust Hub on Aug 1, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill downloads a script from a dynamic, non-vendor subdomain (graceful-tiger-715.convex.site) and executes it using Node.js. This pattern is used to deploy and execute arbitrary code on the local system.
  • [EXTERNAL_DOWNLOADS]: The workflow relies on fetching executable code from an unverified and volatile remote source via curl.
  • [COMMAND_EXECUTION]: The skill executes various commands defined in the output of the external script, including migration gate steps and smoke tests, which are not known beforehand and could be malicious.
  • [PROMPT_INJECTION]: The skill processes untrusted instructions from the check-updates.mjs output, making it vulnerable to indirect prompt injection. 1. Ingestion points: Data and migration steps from the downloaded script (SKILL.md). 2. Boundary markers: None present. 3. Capability inventory: Node.js execution and shell command execution (SKILL.md). 4. Sanitization: None present.
Recommendations
  • HIGH: Downloads and executes remote code from: https://graceful-tiger-715.convex.site/check-updates.mjs - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
CRITICAL
Analyzed
Aug 1, 2026, 05:25 PM
Security Audit — agent-trust-hub — convex-check-updates