settings-management
Pass
Audited by Gen Agent Trust Hub on May 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill documents configuration parameters that execute arbitrary shell commands. Specifically, Claude Code's 'apiKeyHelper', 'hooks', 'statusLine', and 'fileSuggestion' keys, and Codex CLI's 'hooks' block, allow specifying commands that run during the agent's lifecycle.
- [DATA_EXFILTRATION]: Documents OpenCode's variable substitution syntax ('{env:VAR}', '{file:PATH}'), which enables reading sensitive environment variables or local files into the agent context. This poses a risk of secret exposure if exploited via malicious configuration files.
- [EXTERNAL_DOWNLOADS]: Identifies features that trigger remote resource fetching, such as OpenCode's support for npm-based plugins and remote organizational configuration via '.well-known' URLs.
- [PROMPT_INJECTION]: Highlights a significant indirect prompt injection surface by promoting project-level configuration files (e.g., '.claude/settings.json', 'opencode.json') that are committed to version control. This allows malicious repository contributors to alter agent behavior or disable security controls like sandboxing and permission prompts.
Audit Metadata