Issue 1: Validator system prompt is grammatically broken and contradicts itself

[sanjaigridsandguides]

Problem

• The _SYSTEM_PROMPT in nodes/validate_node.py contains lines like:
  • "for two iterations if the confidence of the rsult less than 0.95 go for loop one more time and generate a new answer" — typo "rsult", broken syntax.
  • "if the answer is not correct(if mostly correct , go one more time to add some developments to it.)"
  • "for the first two iterations, if the answer is not perfect set as invalid go for loop one more time"
• These rules contradict the bullet list of validation criteria right above them.

Impact

• The validator LLM gets inconsistent instructions and behaves unpredictably.
• "Confidence < 0.95" is never available inside the validation node — the confidence number lives in the generator's GeneratedAnswer and is not passed through state. The rule is logically dead.
• Looks unprofessional in a production-style codebase.

Required Fix

• Rewrite the validator prompt as a single, coherent rubric.
• Drop the confidence-threshold rule (or actually thread confidence through state if you want to enforce it).
• Run a spell-check / grammar pass over every prompt in the project.