Remove quip/humor guidance from Moonbeam prompts and enforce neutral prompt tone

[Copilot]

Moonbeam prompt definitions included explicit guidance for wit, sarcasm, and joke-like responses. This update removes that behavior from prompt generation so outputs are consistently clear, professional, and direct.

• System prompt tone and behavior cleanup

  • Rewrote MOONBEAM_SYSTEM_INSTRUCTIONS sections (identity, tone, personality, memory_behavior, response_strategy, verification) to remove playful/quip-like directives.
  • Replaced humor-oriented phrasing with neutral, concise, professional language.
  • Preserved existing structure and constraints while changing only tone semantics.

• Redeploy prompt normalization

  • Updated REDPLOY_MOONBEAM_TEXT_PROMPT from “cryptic message” wording to a concise, professional directive.

• Prompt regression guardrails

  • Added packages/backend/src/ai/ai.constants.spec.ts to assert the prompt includes professional wording and excludes prior quip-like phrases.

// packages/backend/src/ai/ai.constants.ts
export const REDPLOY_MOONBEAM_TEXT_PROMPT =
  `Provide a concise, professional message about the future and humanity's role in it.`;

Original prompt

Remove all humoristic or 'quip'-like content from Moonbeam prompts. Go through the codebase and ensure that any prompt generation or output produced by Moonbeam does not include quips, jokes, or playful language. Ensure that prompts are clear, professional, and to-the-point. If quips are embedded in functions, strings, or templates, rewrite them in a neutral, straightforward style. Run tests (if available) to confirm the outputs no longer contain quips.

The following is the prior conversation context from the user's chat exploration (may be truncated):

User: Current user's login: sfreeman422

User: REMOVE ALL QUIPS FROM MOONBEAMS PROMPTS

Assistant: To proceed with your request to remove all quips from Moonbeam's prompts, I need to know which GitHub repository (owner/repo) you'd like the pull request to be created in. Could you provide the repository name?

Assistant: New message from Copilot

This pull request was created from Copilot chat.

[Copilot]

Pull request overview

This PR updates Moonbeam’s prompt constants to remove explicit humor/quip guidance and adds a regression test to keep the system instructions and redeploy prompt phrasing neutral, professional, and direct.

Changes:

• Rewrites sections of MOONBEAM_SYSTEM_INSTRUCTIONS (identity/tone/personality/memory/strategy/verification) to remove wit/sarcasm-oriented directives.
• Updates REDPLOY_MOONBEAM_TEXT_PROMPT from “cryptic” to “concise, professional”.
• Adds a Vitest spec to assert professional wording is present and prior quip-like phrases are absent.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
packages/backend/src/ai/ai.constants.ts	Removes humor-oriented tone guidance and updates the redeploy text prompt to professional wording.
packages/backend/src/ai/ai.constants.spec.ts	Adds regression tests to prevent reintroduction of quip-like phrasing in the prompt constants.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.