lwy-self
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill manages personal data exclusively in local directories (primarily
~/.learnwy/). All script logic for indexing, linting, and logging is performed locally with no detected network exfiltration or remote communication. - [COMMAND_EXECUTION]: The provided
scripts/cli.cjsallows the agent to maintain the personal store. It includes installation functionality that modifies IDE configuration files (e.g.,~/.claude/settings.json,~/.trae/hooks.json) to register the skill's automated hooks, which is standard behavior for extending agent capabilities. - [PROMPT_INJECTION]: The skill facilitates the ingestion of external data, specifically meeting transcripts and chat logs from Feishu/Lark platforms. This creates a surface for indirect prompt injection if those external sources contain adversarial content.
- Ingestion points: Meeting transcripts pulled from Lark and stored in
raw/lark/before being compiled intowiki/pages. - Boundary markers: None identified in the provided scripts or instructions to isolate external data from the agent's instructions.
- Capability inventory: Local file system access and configuration management via Node.js scripts.
- Sanitization: No specific sanitization or filtering of external transcript content is implemented.
Audit Metadata