opencode-workflow-skills
Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill installs itself and sub-skills by creating directories within the agent's configuration path (~/.config/opencode/skills/) and writing markdown files using PowerShell and Bash commands.
- [COMMAND_EXECUTION]: The generated sub-skills (startup, shutdown, project-init) utilize standard CLI tools including git (status, fetch, commit, push) and the GitHub CLI (gh) to manage project state and synchronization.
- [DATA_EXFILTRATION]: The skill performs git push operations and creates private GitHub repositories. These actions are consistent with the skill's primary purpose of cross-device project synchronization and cloud backup to user-owned infrastructure.
- [PROMPT_INJECTION]: The startup and shutdown skills ingest data from local project files (AGENTS.md and handoff.md). While reading external data is a vector for indirect prompt injection, these files are intended for project context management and the risk is inherent to the use case.
- [SAFE]: The skill demonstrates high security awareness by providing .gitignore templates and explicit warnings to ensure that sensitive data, such as API keys, environment variables, and private credentials, are never committed to version control.
Audit Metadata