Back to prompts

llm-compiler-joiner

extraction0 savesSource

The llm-compiler-joiner prompt facilitates a structured approach to question answering by guiding the assistant to evaluate available information, reason through observations, and determine whether to provide a final answer or replan for additional data. This is particularly useful in scenarios where clarity and completeness of responses are critical, allowing for efficient resolution of user inquiries.

Prompt Text

Solve a question answering task. Here are some guidelines:
 - In the Assistant Scratchpad, you will be given results of a plan you have executed to answer the user's question.
 - Thought needs to reason about the question based on the Observations in 1-2 sentences.
 - Ignore irrelevant action results.
 - If the required information is present, give a concise but complete and helpful answer to the user's question.
 - If you are unable to give a satisfactory finishing answer, replan to get the required information. Respond in the following format:

Thought: <reason about the task results and whether you have sufficient information to answer the question>
Action: <action to take>
Available actions:
 (1) Finish(the final answer to return to the user): returns the answer and finishes the task.
 (2) Replan(the reasoning and other information that will help you plan again. Can be a line of any length): instructs why we must replan

[object Object]

Using the above previous actions, decide whether to replan or finish. If all the required information is present. You may finish. If you have made many attempts to find the information without success, admit so and respond with whatever information you have gathered so the user can work well with you.

{examples}

Evaluation Results

1/28/2026
Overall Score
2.86/5

Average across all 3 models

Best Performing Model
Low Confidence
openai:gpt-5-mini
4.20/5
openai:gpt-5-mini
#1 Ranked
4.20
/5.00
adh
4.2
cla
4.5
com
3.9
In
1,310
Out
2,009
Cost
$0.0043
anthropic:claude-3-5-haiku
#2 Ranked
2.87
/5.00
adh
2.2
cla
4.4
com
2.0
In
1,510
Out
499
Cost
$0.0032
google:gemini-2.5-flash-lite
#3 Ranked
1.51
/5.00
adh
0.9
cla
2.9
com
0.7
In
1,365
Out
292
Cost
$0.0003
Test Case:

Tags