opencode-acp-control

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION] (CRITICAL): The skill documentation suggests a manual update path using 'curl -fsSL https://opencode.dev/install | bash'. This piped-to-shell pattern from an untrusted domain allows for arbitrary remote code execution on the user's system.
  • [COMMAND_EXECUTION] (HIGH): The skill relies on the bash tool for core functions like starting sessions and checking versions. While functional, this provides a wide attack surface for command injection if inputs are not properly sanitized.
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill fetches version data from github.com and suggests an installer from opencode.dev, which is not on the list of authorized trusted sources.
  • [Indirect Prompt Injection] (LOW): 1. Ingestion points: The skill reads output from the opencode process via process.poll and web content via webfetch (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Includes bash command execution, process.write, and process.kill (SKILL.md). 4. Sanitization: No evidence of sanitization of external data is provided before it influences agent logic.
Recommendations
  • HIGH: Downloads and executes remote code from: https://opencode.dev/install - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
CRITICAL
Analyzed
Feb 17, 2026, 06:31 PM