An absent (or None) classifier_model only says the run carries no preference, so the classifier keeps whatever the server resolved at startup (--auto-classifier-model, DEEPAGENTS_CODE_AUTO_CLASSIFIER_MODEL, [models].auto_classifier). /auto model clear needs the stronger statement that reviews go back to the main agent model, which this sentinel carries.

It cannot collide with a real spec: create_model resolves provider:model (or a bare model name) and has no provider or model named __dcode_.... A control character such as a leading NUL would also be collision-proof, but this value has to survive the trip to a remote deployment intact — the context is serialized to JSON and may be persisted, and Postgres text/jsonb rejects NUL outright. A stripped sentinel would silently read as "no preference" and leave a startup classifier authorizing actions after the UI reported the clear, so the sentinel stays plain ASCII.

attribute

AUTO_CLASSIFIER_TIMEOUT_SECONDS_DEFAULT: float

Default wall-clock budget for one Auto classifier decision batch.

Single source of truth shared by the manifest option, the middleware default, and the resolver, so the three cannot drift (pinned by test).

attribute

logger

attribute

AUTO_MODE_COUNTERS_NAMESPACE: tuple[str, str]

attribute

USER_PROMPT_METADATA_KEY: str

attribute

AUTO_MODE_EVENT_TYPE: str

attribute

DecisionDisposition: Literal['deterministic_allow', 'classifier_allow', 'policy_deny', 'classifier_unavailable', 'require_human']

Functions

function

approval_mode_key

Return the store key for a thread's live approval mode.

function

aread_approval_mode_from_store

Asynchronously read a live approval mode from a LangGraph Store.

The graph server supplies an async batched Store whose synchronous methods reject calls from the event-loop thread. Prefer aget() for that runtime, while retaining a synchronous fallback for lightweight local test stores.

function

coerce_approval_mode

Return a validated mode, failing closed to manual.

function

project_goal_state

Project authoritative channels into deterministic notice state.

function

user_prompt_metadata

Build trusted classifier metadata for a client-created user message.

function

mcp_tool_is_coherently_read_only

Return whether an MCP tool has coherent read-only annotations.

function

is_mcp_tool

Return whether a tool carries dcode's MCP wrapper marker.

function

gated_mcp_tool_names

Return MCP names that require Manual or Auto review.

function

sanitize_auto_reason

Return a compact reason safe for persistence, logs, and UI rendering.

function

classifier_unavailable_reason

Build a safe agent/UI reason for a failed auto classifier call.

Provider exception text stays out of the reason (it can carry secrets or noisy HTML). Only real local deadline expiry (_ClassifierDeadlineExceededError) says the classifier did not respond within the configured wait budget; a bare provider TimeoutError stays type-only so we do not claim dcode's deadline fired when the model failed first. A configured classifier model that cannot be built is named as such so the user fixes the setting instead of waiting out a nonexistent outage.