opencode-build-plugins

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill contains an implementation pattern in references/update-notifications.md for checking plugin updates by fetching metadata from the npm registry (registry.npmjs.org). This is a well-known service used for its intended purpose of version discovery.
  • [COMMAND_EXECUTION]: The skill provides code examples in references/examples.md and references/hook-patterns.md that use the Bun Shell ($) to execute system commands such as osascript (for macOS notifications), wc (word count), and eslint. These are presented as standard functional templates for plugin developers.
  • [INDIRECT_PROMPT_INJECTION]: The skill acts as a developer assistant that generates TypeScript plugin code based on user-supplied requirements. This creates an indirect prompt injection surface where malicious instructions in a user prompt could influence the generated plugin logic.
  • Ingestion points: User prompts describing desired plugin functionality.
  • Boundary markers: The instructions do not define specific delimiters or warnings for user-provided instructions.
  • Capability inventory: Generated plugins have the capability to intercept tool calls, execute shell commands, and perform network requests.
  • Sanitization: The skill emphasizes type safety and modularity but does not prescribe specific sanitization for the inputs used to generate the plugin logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 06:04 AM