dln
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script (
ks-merge.py) located within its package structure to synchronize Knowledge State data between the agent session and the Notion database. This script is used to perform precise data merging using structured JSON payloads. - [PROMPT_INJECTION]: The skill manages potential indirect prompt injection risks by implementing HTML comment boundary markers (
<!-- KS:start -->and<!-- KS:end -->) to isolate processed data from external sources. This ensures that the agent correctly distinguishes between learner-specific data and system instructions. - [SAFE]: No malicious patterns such as data exfiltration to untrusted domains, credential harvesting, or unauthorized remote code execution were identified. The skill follows best practices for tool integration and state persistence.
Audit Metadata