20 of the best prompts for diagnosing custom instruction ignorance, step by step across 4 stages. Works with ChatGPT, Claude, and Gemini.
ChatGPT sometimes fails to adhere to custom instructions set in the interface, leading to responses that don't align with user expectations. This can result in confusion or frustration as users receive irrelevant or off-topic answers. By following this guide, users can effectively diagnose, fix, and prevent this issue from recurring in future interactions. Built across 4 distinct stages covering Diagnose the Wall, Isolate and Fix, Verify the Resolution and more, this guide gives you one expert prompt per step so you never have to write from scratch or guess what the AI needs. The prompts work in ChatGPT, Claude, and Gemini and are designed to get usable output on the first try.
Stage 1
Identifying the specific failure in ChatGPT's adherence to custom instructions is crucial before attempting any fixes.
Identify the Ignored Instruction
Identify the specific custom instruction that was not followed in your last response: [INSERT SPECIFIC INSTRUCTION]. Quote the exact part of your previous response that deviated from this instruction and explain how it differs from what was requested.
Analyze the Failed Output
Paste the last response you provided and identify the section where you did not follow my custom instruction about [INSERT SPECIFIC INSTRUCTION]. Analyze the reasons for this non-compliance and suggest how you could improve your adherence to the instruction in future responses.
Explain the Mechanism of Ignorance
What processing mechanism led you to overlook my custom instruction regarding [INSERT SPECIFIC INSTRUCTION]? Please analyze your response and explain the reasoning that contributed to this oversight, including any specific factors that may have influenced your interpretation.
Compare Expected vs. Actual Output
Compare the expected output based on my custom instruction, which is [INSERT EXPECTED OUTPUT], with your actual output, [INSERT ACTUAL OUTPUT]. Identify and explain the discrepancies between these two outputs, highlighting any specific areas where the actual output did not align with my expectations.
Identify the Root Cause
Identify the specific limitation or behavior pattern that caused you to ignore my custom instruction about [INSERT SPECIFIC INSTRUCTION]. Please provide a detailed explanation of this failure and categorize it into one of the following types: misunderstanding, misinterpretation, or technical limitation.
Stage 2
To correct the failure of ignoring custom instructions, we need to apply specific fixes.
Re-attempt with Instruction Reminder
Restate the following custom instruction before providing your response: [INSERT SPECIFIC INSTRUCTION]. After restating, ensure your answer aligns with the instruction and addresses the question or topic at hand.
State Correct Behavior Before Answering
Confirm that you will adhere to my custom instruction about [INSERT SPECIFIC INSTRUCTION] before providing your answer. After confirming, please respond to my question with the necessary details and clarity.
Break Down the Task
Summarize the following custom instruction: [INSERT SPECIFIC INSTRUCTION]. Then, outline a step-by-step response that demonstrates how to effectively follow this instruction, ensuring clarity and adherence to the specified guidelines.
Enforce Output Format
Format your answer according to the following structure: [INSERT DESIRED FORMAT]. Additionally, make sure to adhere to my custom instruction about [INSERT SPECIFIC INSTRUCTION] to ensure clarity and consistency in your response.
Role Framing for Compliance
Assume the role of a strict adherence assistant. Your task is to follow the custom instruction provided: [INSERT SPECIFIC INSTRUCTION]. Explain how you will ensure compliance with this instruction in your response, and provide an example of how you would apply it in a relevant scenario.
Stage 3
Testing the effectiveness of the fixes will confirm whether the issue has been resolved.
Replay the Original Request
Repeat the original request I made that was not followed according to my custom instruction: [INSERT ORIGINAL REQUEST]. After that, provide a response that adheres to the custom instruction, ensuring clarity and relevance to the original request.
Test with a Variation
Test the following variation of the original request: [INSERT VARIATION]. Confirm that you adhere to my custom instruction regarding [INSERT SPECIFIC INSTRUCTION] and provide a detailed explanation of how you incorporated it into your response.
Self-Check Before Submission
Self-check your response against the following custom instruction: [INSERT SPECIFIC INSTRUCTION]. After reviewing, confirm whether your answer complies with this instruction and explain any areas where it may not fully align.
Compare Outputs
Generate two versions of your response: one that follows my custom instruction [INSERT SPECIFIC INSTRUCTION] and another that does not adhere to it. Analyze and identify the key differences between the two responses, highlighting how the instruction influenced the content and tone.
Flag Uncertainty Areas
Flag any areas in your response where you felt uncertain about following my custom instruction: [INSERT SPECIFIC INSTRUCTION]. For each flagged area, provide a brief explanation of why you felt uncertain and suggest how you could better align with the instruction in future responses.
Stage 4
Creating reusable assets will help prevent this failure from happening again.
Custom Instruction Reminder Template
Remind yourself before generating a response to adhere to the custom instruction: [INSERT SPECIFIC INSTRUCTION]. If you are uncertain about any aspect of the instruction, please clarify your understanding before proceeding with the response.
Output Format Constraint
Format your responses according to the following structure: [INSERT DESIRED FORMAT]. Ensure that your output strictly adheres to the custom instruction: [INSERT SPECIFIC INSTRUCTION], and provide an example of how this format looks in practice.
Self-Check Protocol
Implement a self-check protocol where you confirm your understanding of and commitment to following the custom instruction: [INSERT SPECIFIC INSTRUCTION]. After this confirmation, provide a brief summary of the instruction to ensure clarity before proceeding with the response.
Instruction Compliance Trigger Phrase
Start each session by stating: "I will strictly follow the custom instruction: [INSERT SPECIFIC INSTRUCTION]." Then, confirm understanding of the instruction and provide a brief summary of how you will apply it throughout the session.
Session Opener for Instruction Adherence
At the beginning of each session, state clearly: 'Today, I will ensure compliance with the custom instruction: [INSERT SPECIFIC INSTRUCTION].' Then, provide a brief explanation of how you plan to adhere to this instruction throughout the session, including specific examples or guidelines to follow.
ChatGPT may ignore custom instructions due to its reliance on context and the potential for conflicting information in its training data. This can lead to prioritizing general knowledge over specific user directives. Additionally, if the context window is exceeded, earlier instructions may be lost.
To enhance adherence to your instructions, clearly state them at the beginning of each interaction. ChatGPT's context window can limit its ability to recall earlier instructions, so restating them can help maintain focus. Using specific language and formatting can also guide its responses more effectively.
If ChatGPT deviates from your instructions, provide feedback by quoting the specific part of its response that was incorrect. This helps the model understand where it went wrong and adjust its future responses. Remember that its training data may influence its interpretation of instructions.
ChatGPT does not retain memory of instructions across sessions unless using a paid plan with memory features. Each new session starts without prior context, so it's essential to restate your instructions every time. This ensures that the model aligns its responses with your expectations.
ChatGPT's limitations include its context window, which can lead to forgetting earlier instructions if the conversation becomes lengthy. Additionally, its training data may not always align with specific user instructions, causing it to prioritize general knowledge. Understanding these limitations can help users frame their requests more effectively.
AI Prompts for Identifying Ignored Custom Instructions
Claude sometimes fails to follow custom instructions or system prompts set in the interface, leading to responses that do not align with user expectations.
See promptsAI Prompts for Identify Custom Instruction Failures
When Gemini ignores custom instructions, it can lead to responses that don't align with user expectations.
See promptsAI Prompts for Identify Ignored Custom Instructions
Grok sometimes fails to adhere to custom instructions or system prompts, leading to responses that do not align with user expectations.
See prompts