lexiang-knowledge-base
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements robust security guardrails, including mandatory rules that prohibit the agent from guessing document IDs and require explicit user approval for any write, modification, or deletion operations.
- [CREDENTIALS_UNSAFE]: Authentication is correctly managed through configuration templates in mcp.json using environment-style placeholders for tokens. No hardcoded credentials, API keys, or secrets were detected within the instructions or scripts.
- [EXTERNAL_DOWNLOADS]: All external URLs and API endpoints refer to official vendor domains (lexiangla.com and lexiang-app.com), which are trusted sources for the skill's operations. The instructions guide users to official platforms for configuration.
- [COMMAND_EXECUTION]: The skill uses standard command-line tools like curl for binary file uploads to pre-signed URLs. This workflow is transparently documented and requires the agent to provide the command to the user for execution, maintaining transparency and user control.
- [REMOTE_CODE_EXECUTION]: The provided helper scripts (scripts/upload-files.py and scripts/sync-folder.ts) are transparent utility tools for local file management. They do not implement dynamic code execution from untrusted sources or malicious remote script piping.
- [PROMPT_INJECTION]: The instructions utilize mandatory instructional language to enforce safety boundaries and routing logic rather than attempting to bypass safety filters or override the agent's core behavioral constraints.
Audit Metadata