agentscope-ai / agentscope-ai/agentscope
[Feature]: Sandbox mechanism during skill runtime
Aberta
- Linguagem predominante
- Python
- Estrelas
- 31.6k
- Forks
- 3.5k
- Merge médio
- 1d 16h
- PRs com merge (30d)
- 103
Descrição
In agentScope, regarding the runtime of skills and the use of capabilities such as execute_shell_command, execute_python_code, and view_text_file, should a unified sandbox be provided to restrict the dynamic nature of these LLM calls?
Guia de contribuição
Avaliação
Esta issue ainda não foi avaliada.