lovstudio-cc-mv
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads and executes the
@lovstudio/cc-mvpackage from the NPM registry usingnpxto perform its core functions. - [COMMAND_EXECUTION]: It invokes shell commands through
npxto preview and execute migration tasks, incorporating file paths and session identifiers derived from user input. - [DATA_EXFILTRATION]: The skill accesses and modifies internal Claude Code state files, including
~/.claude/projects/,~/.claude/history.jsonl, and~/.claude/sessions/. This access is necessary for the intended purpose of synchronizing history with moved project directories. - [PROMPT_INJECTION]: The skill processes untrusted user data (file paths and session descriptions) and interpolates it into shell command templates, which represents a potential attack surface.
- Ingestion points: User-provided directory paths and session content descriptions (SKILL.md, Step 1 and Step 3).
- Boundary markers: No explicit delimiters are used for the path interpolation in the command execution steps.
- Capability inventory: Execution of shell commands via
npxand direct filesystem manipulation. - Sanitization: The instructions direct the agent to resolve paths to absolute locations and expand home directory symbols (~), providing a basic level of input validation.
Audit Metadata