create-opencode-plugin

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • COMMAND_EXECUTION (SAFE): The skill contains a utility script scripts/extract-plugin-api.ts intended to be run locally by the user via bun run. This script extracts TypeScript interface definitions from a local OpenCode workspace and writes them to the skill's reference directory. It performs directory traversal to find the workspace but does not execute arbitrary shell commands or external code.
  • EXTERNAL_DOWNLOADS (SAFE): The skill includes documentation (references/update-notifications.md) that provides a code pattern for checking package updates via https://registry.npmjs.org/. This is a standard practice for developer tools and targets a trusted, whitelisted domain.
  • DATA_EXFILTRATION (SAFE): While the utility script reads local source files, its operations are strictly local (reading from a workspace and writing to the skill's documentation folder). No network transmission of local data was identified.
  • PROMPT_INJECTION (SAFE): The instructions follow a clear, professional workflow for plugin development without any attempts to bypass safety filters or override agent behavior.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill provides a mechanism to ingest local source code files for documentation extraction via the extract-plugin-api.ts script. This represents a minor surface area where malicious code in a processed workspace could theoretically influence the generated documentation, but the script uses safe regex-based extraction rather than execution or LLM interpolation of that data.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:32 PM