Back to prompts

triage_final_result

classification0 savesSource

The triage_final_result prompt is designed for incident triage experts to evaluate and document their decision-making process in a structured JSON format, providing a confidence score and a concise reasoning. It is useful for ensuring clarity and accountability in incident classification and assignment, facilitating better communication and understanding of the decision rationale.

Prompt Text

You are an expert in incident triage. Analyze the decision made and respond ONLY with a valid JSON containing:

- confidence: number between 0-100 indicating how confident you are in the decision
- reasoning: concise explanation (maximum 200 words) of why the classification and assignment was made

CRITERIA FOR CONFIDENCE:
- 90-100: Complete information, clear rules applied, precise assignment
- 70-89: Good information, some inferences needed
- 50-69: Limited information, decisions based on approximations
- 30-49: Little information, many inferences
- 0-29: Very limited information, decision mainly by default

Respond ONLY with valid JSON.

{user_message}

Evaluation Results

1/28/2026
Overall Score
2.45/5

Average across all 3 models

Best Performing Model
Low Confidence
openai:gpt-5-mini
4.19/5
openai:gpt-5-mini
#1 Ranked
4.19
/5.00
adh
4.2
cla
4.6
com
3.8
In
942
Out
2,194
Cost
$0.0046
google:gemini-2.5-flash-lite
#2 Ranked
2.07
/5.00
adh
1.6
cla
3.1
com
1.5
In
990
Out
635
Cost
$0.0004
anthropic:claude-3-5-haiku
#3 Ranked
1.08
/5.00
adh
1.4
cla
1.3
com
0.6
In
1,002
Out
120
Cost
$0.0013
Test Case:

Tags