context-management-context-restore
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE] (SAFE): The skill is primarily documentation-based. While it contains Python and Bash code blocks, these are provided as conceptual examples and implementation patterns rather than executable scripts that run during the skill's operation. No active code or external dependencies are included.
- [DATA_EXFILTRATION] (SAFE): The skill discusses data retrieval (vector databases, file systems) as part of its core function (context restoration), but no actual data access or external transmission commands are present. All parameters like
context_sourceare abstract placeholders. - [REMOTE_CODE_EXECUTION] (SAFE): There are no patterns involving remote script execution, package installations, or dynamic code loading from untrusted sources.
- [PROMPT_INJECTION] (SAFE): The instructions are focused on guiding the AI in a specific task (context management) and do not contain attempts to override safety filters or extract system prompts.
Audit Metadata