DatabackendOptions

Options that control which backend fetch and commit operations are performed during ChatObject execution.

Description

DatabackendOptions provides fine-grained control over the framework-managed fetch strategy. By setting flags to True, you can skip specific backend operations, which is useful for performance optimization or when certain data is already available.

Fields

• skip_memory_fetch (bool): Skip loading memory from the backend (default: False)
• skip_tools_fetch (bool): Skip loading tools from the backend (default: False)
• skip_mcp_fetch (bool): Skip loading MCP clients from the backend (default: False)
• skip_presets_fetch (bool): Skip loading presets from the backend (default: False)
• skip_ability_extra_setting (bool): Skip loading extra ability settings from the backend (default: False)
• skip_memory_commit (bool): Skip committing memory back to the backend after execution (default: False)

Usage

v0.12.0 migration: DatabackendOptions has been moved from amrita_core.chatmanager.chat_object to amrita_core.contexts. The old import path still works via backward-compatible re-exports, but updating to the new path is recommended.

from amrita_core.contexts import DatabackendOptions

# Only load memory, skip everything else
opts = DatabackendOptions(
    skip_tools_fetch=True,
    skip_mcp_fetch=True,
    skip_presets_fetch=True,
    skip_ability_extra_setting=True,
)

chat = ChatObject(
    train=train,
    user_input="Hello",
    session_id="session_123",
    backend_options=opts,
)