manual-interacting-with-opencode-via-cli

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is a comprehensive guide for manual verification and testing using the OpenCode CLI, emphasizing security-positive patterns such as process isolation and secret redaction.
  • [EXTERNAL_DOWNLOADS]: The instructions refer to the installation of the opencode-ai package via NPM, which is the expected tool for the skill's functionality.
  • [COMMAND_EXECUTION]: The skill directs the agent to execute CLI commands using isolated environments, specifically recommending temporary HOME and XDG directories to prevent interference with the host system.
  • [PROMPT_INJECTION]: The skill handles user-provided tasks through the opencode run command, representing an indirect prompt injection surface.
  • Ingestion points: User-defined tasks passed to the CLI.
  • Boundary markers: Tasks are encapsulated within placeholders (e.g., <task>).
  • Capability inventory: The agent executes the opencode binary which has capabilities for file system and potentially network interaction.
  • Sanitization: The skill mandates strict environment isolation and warns against printing sensitive values to the terminal.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 08:59 PM
Security Audit — agent-trust-hub — manual-interacting-with-opencode-via-cli