lovstudio-auto-context
Pass
Audited by Gen Agent Trust Hub on May 7, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill performs file system modifications to maintain project-specific memory and update agent configuration files. It interacts with local paths such as
~/.claude/projects/and both global and project-levelCLAUDE.mdfiles to preserve session context and rules. This behavior is aligned with the skill's stated purpose as a context operator. - [PROMPT_INJECTION]: The skill monitors specific linguistic patterns in the session transcript (e.g., '从今以后', '记住') to identify instructions or preferences that should be persisted to memory. While this involves processing untrusted user input, it is a functional feature for context hygiene and does not attempt to bypass or override agent safety protocols.
- [SAFE]: Modifications to sensitive configuration files like
~/.claude/CLAUDE.mdare governed by a mandatory 'Confirm-first' flow. The skill instructions require the agent to display a diff preview and obtain explicit user consent before applying changes, which mitigates the risk of silent or unauthorized configuration updates.
Audit Metadata