Issue — Add confidence_tier computed property to ReasoningResult

[Devnil434]

Description: The dashboard needs to colour-code alerts by confidence level but the frontend shouldn't duplicate the threshold logic. Add a confidence_tier computed property to ReasoningResult:

@property
def confidence_tier(self) -> Literal["high", "medium", "low"]:
    if self.confidence >= 0.75:  return "high"
    if self.confidence >= 0.50:  return "medium"
    return "low"

And a is_actionable property: label == "Suspicious" and confidence >= 0.65.

Files to touch:

• libs/schemas/reasoning.py
• tests/test_reasoning_schema.py (create)

Acceptance criteria:

• 6 unit tests covering all tier boundaries (0.49, 0.50, 0.74, 0.75, 1.0)
• is_actionable tested: True only when label = Suspicious AND conf ≥ 0.65
• Both properties serialised correctly when model_dump(mode="json") is called

---

---

[Prarthana124]

Hi @Devnil434, I would like to work on this issue under GSSoC 2026.

I checked the current implementation in libs/schemas/reasoning.py and noticed that the confidence_tier and is_actionable properties are already present. My approach would be to verify and improve the serialization behaviour so both computed properties are correctly included when model_dump(mode="json") is called.

I will create tests/test_reasoning_schema.py and add the required boundary-condition tests for the specified confidence values (0.49, 0.50, 0.74, 0.75, and 1.0) to ensure the classification logic behaves correctly across edge cases. I will also verify the is_actionable behaviour for different label/confidence combinations while keeping the existing schema behaviour unchanged.

Please assign this issue to me.

[ArokyaMatthew]

@Devnil434 , I would like to work on this issue as a part of GSSoC