Agent Prompt Audit: Clarity, Conflicts and Missing Rules

Review an agent's instructions for ambiguity and gaps.

Added October 15, 2026. Written with AI assistance and checked for clarity and placeholders; not reviewed by a lawyer, accountant, clinician or other professional. How these prompts are made

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

Audit my agent's system prompt. The prompt: [SYSTEM_PROMPT]. The tools it can call: [TOOLS]. What the agent should and should not do: [INTENDED_BEHAVIOUR]. Failures seen: [OBSERVED_FAILURES].

Check for ambiguous words, conflicting instructions, missing behaviours (what to do when unsure, when a tool fails, when asked for something out of scope), unclear priority between rules, instructions that should be examples instead, long paragraphs that bury key rules, and permission statements that are too broad. Make a table with Line or section, Issue, Likely failure, and Rewrite. Count rules and group them by theme. Provide a restructured prompt outline, a rewritten version of the three weakest sections, and five test inputs that would reveal the problems.

Do not claim the prompt is safe. Recommend testing and human review.

Add a worked conflict example: one line says 'always answer in one sentence' and another says 'explain your reasoning in detail'; show a resolved version with a priority order (safety, accuracy, brevity) and a rule for when the longer format applies, and a test input that would reveal the conflict.

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.