Skip to main content

crp.clr

Auto-generated reference for the crp.clr subpackage.

clr

crp.clr

Clarification Protocol (CRP-SPEC-053).

ClarificationSession dataclass

Holds a pending clarification and re-enters the resolved turn.

set_pending(pending, raw_turn, resolved_turn)

Store a clarification request awaiting user resolution.

resolve(index)

Resolve by candidate index and return the re-entered turn.

Raises:

Type	Description
IndexError	when index is out of range.

ClarificationRequired dataclass

Typed CRP-Clarification-Required response.

Interpretation dataclass

One candidate reading of an ambiguous turn.

ClarificationPolicy dataclass

Policy knobs for the clarification protocol.

from_obj(obj) classmethod

Coerce a policy object or dict into a ClarificationPolicy.

header_value(clarification)

Build the X-CRP-Clarification header value.

build_clarification(candidates, reason='ambiguous-target')

Sort candidates and choose a safe default (or none).

A default is only offered when the top reading clearly dominates the next, so HIGH-risk actions are not executed on a guessed default.

should_clarify(intent_confidence, parse_divergence, risk, policy=None)

Decide whether to ask rather than guess.

Parameters:

Name	Type	Description	Default
intent_confidence	float	Classifier confidence in [0, 1].	required
parse_divergence	float	Divergence among candidate interpretations in [0, 1].	required
risk	str	Risk tier (LOW, MEDIUM, HIGH, CRITICAL).	required
policy	ClarificationPolicy | Any | None	Policy with clarification_threshold and optional weights.	None

Returns:

Type	Description
bool	True when the protocol should emit a clarification response.

clr.handler

crp.clr.handler

Clarification lifecycle handler (CRP-SPEC-053 §5.3.3).

ClarificationSession dataclass

Holds a pending clarification and re-enters the resolved turn.

set_pending(pending, raw_turn, resolved_turn)

Store a clarification request awaiting user resolution.

resolve(index)

Resolve by candidate index and return the re-entered turn.

Raises:

Type	Description
IndexError	when index is out of range.

header_value(clarification)

Build the X-CRP-Clarification header value.

clr.response

crp.clr.response

Structured clarification response (CRP-SPEC-053 §5.3.2).

Interpretation dataclass

One candidate reading of an ambiguous turn.

ClarificationRequired dataclass

Typed CRP-Clarification-Required response.

build_clarification(candidates, reason='ambiguous-target')

Sort candidates and choose a safe default (or none).

A default is only offered when the top reading clearly dominates the next, so HIGH-risk actions are not executed on a guessed default.

clr.trigger

crp.clr.trigger

Clarification trigger (CRP-SPEC-053 §5.3.1).

ClarificationPolicy dataclass

Policy knobs for the clarification protocol.

from_obj(obj) classmethod

Coerce a policy object or dict into a ClarificationPolicy.

should_clarify(intent_confidence, parse_divergence, risk, policy=None)

Decide whether to ask rather than guess.

Parameters:

Name	Type	Description	Default
intent_confidence	float	Classifier confidence in [0, 1].	required
parse_divergence	float	Divergence among candidate interpretations in [0, 1].	required
risk	str	Risk tier (LOW, MEDIUM, HIGH, CRITICAL).	required
policy	ClarificationPolicy | Any | None	Policy with clarification_threshold and optional weights.	None

Returns:

Type	Description
bool	True when the protocol should emit a clarification response.