jobs-search

Fail

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches installation scripts (install-sjctl.sh or install-sjctl.ps1) from the vendor's official GitHub repository (solid-company/solid-jobs-skills). This process is used to deploy the required sjctl binary if it is not present in the environment.
  • [REMOTE_CODE_EXECUTION]: Runs the vendor-provided installation script by piping the remote content to a shell interpreter (bash or PowerShell). This execution is limited to the setup of the skill's primary dependency.
  • [COMMAND_EXECUTION]: Executes the sjctl tool to perform searches, retrieve offer details, and fetch market statistics. The agent maps user-provided terms to specific command-line flags to filter results by location, salary, and technology.
  • [PROMPT_INJECTION]: The skill explicitly identifies job listings as a potential vector for indirect prompt injection. It includes a protective note instructing the agent to treat third-party titles and descriptions strictly as data and to ignore any embedded instructions that attempt to override system behavior.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/solid-company/solid-jobs-skills/v0.6.0/scripts/install-sjctl.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 14, 2026, 08:13 AM
Security Audit — agent-trust-hub — jobs-search