The Agent Skills Directory

[PROMPT_INJECTION]: No malicious instructions or bypass attempts were detected. The skill instructions are focused entirely on project scaffolding and conventions.\n- [DATA_EXFILTRATION]: The skill does not perform any network operations or access sensitive data. It operates strictly within the project's laboratory directory.\n- [REMOTE_CODE_EXECUTION]: No remote code execution or untrusted package installations are present. The skill provides static templates for local file creation.\n- [COMMAND_EXECUTION]: The use of 'git mv' is appropriate for the task of migrating existing experiment files into the new laboratory structure and does not represent a security risk.\n- [SAFE]: The skill implements safety measures such as 'noindex' robots metadata and correctly scopes filesystem operations to the laboratory directory using standard Next.js patterns.

build-lab