run_textual_cli_async(
assistant_id: str,
*,
approval_mode: ApprovalMode | Name | Type | Description |
|---|---|---|
assistant_id* | str | Agent identifier for memory storage. |
approval_mode | ApprovalMode | str | Default: 'manual'Initial |
auto_approve | bool | None | Default: None |
sandbox_type | str | Default: 'none' |
sandbox_id | str | None | Default: None |
sandbox_snapshot_name | str | None | Default: None |
sandbox_setup | str | None | Default: None |
model_name | str | None | Default: None |
model_params | dict[str, Any] | None | Default: None |
profile_override | dict[str, Any] | None | Default: None |
thread_id | str | None | Default: None |
resume_thread | str | None | Default: None |
initial_prompt | str | None | Default: None |
initial_skill | str | None | Default: None |
initial_goal | str | None | Default: None |
startup_cmd | str | None | Default: None |
mcp_config_path | str | None | Default: None |
no_mcp | bool | Default: False |
trust_project_mcp | bool | None | Default: None |
hook_trust | WorkspaceTrust | None | Default: None |
enable_interpreter | bool | None | Default: None |
interpreter_arg | bool | None | Default: None |
interpreter_ptc | str | list[str] | None | Default: None |
interpreter_ptc_acknowledge_unsafe | bool | Default: False |
allow_fs_tools | list[FsToolName] | None | Default: None |
auto_classifier_model | str | None | Default: None |
recursion_limit | int | None | Default: None |
Run the Textual TUI interface (async version).
Starts a LangGraph server in a subprocess and connects the TUI to it via the
langgraph-sdk client.
Compatibility input for callers using the previous Boolean
API. True maps to unrestricted yolo.
Type of sandbox ("none", "agentcore", "modal", "runloop", "daytona", "langsmith")
Optional existing sandbox ID to reuse.
Snapshot (langsmith) or blueprint (runloop) name.
Optional path to setup script to run in the sandbox after creation.
Optional model name to use
Extra kwargs from --model-params to pass to the model.
These override config file values.
Extra profile fields from --profile-override.
Merged on top of config file profile overrides.
Thread ID for the session.
None when resume_thread is provided (the TUI resolves the final
ID asynchronously).
Raw resume intent from -r flag.
'__MOST_RECENT__' for bare -r, a thread ID string for -r <id>,
or None for new sessions.
Resolved asynchronously inside the TUI.
Optional prompt to auto-submit when session starts
Optional skill name to invoke when the session starts.
Optional goal objective to draft criteria for when the session starts.
Shell command to run at startup before the first prompt.
Output is rendered in the transcript; non-zero exits warn but do not abort the session.
Optional path to MCP servers JSON configuration file.
Merged on top of auto-discovered configs (highest precedence).
Disable all MCP tool loading.
Controls project-level server trust (stdio and remote alike).
True to allow, False to deny, None to fall back to the
user's per-server scoped approvals (equivalent to False for the
whole-config decision).
Policy deciding which workspaces may run project-scoped hook
commands. None consults only the persisted trust store.
Enable CodeInterpreterMiddleware (js_eval) on
the main agent. None defers to the sandbox-aware/config default.
The raw --interpreter/--no-interpreter tri-state,
forwarded so the app can tell an explicit opt-out from a
sandbox-suppressed default when surfacing the disabled-by-sandbox
advisory.
Override for settings.interpreter_ptc (PTC allowlist
for js_eval).
Explicit acknowledgement for
interpreter_ptc="all" outside of auto_approve.
Allowlist for FilesystemMiddleware's tools param,
from --allow-fs-tools.
None leaves the SDK default (all tools).
Model spec the Auto approval classifier reviews
with, from --auto-classifier-model.
None resolves from env / config.toml and then reuses the main
agent model. A blank string means the flag was explicitly supplied
with no value: it overrides any env / config.toml classifier so
reviews inherit the main agent model.
Explicit main-agent recursion_limit; None resolves
from env / config.toml / default at agent-build time.