Is it possible to write an additional system prompt to make it less agreeable

[woss]

Question

Don't get me wrong but i personally don't need a validation from the AI 😊

Is it possible to write a custom prompt like that would be cold as ice?

[rekram1-node]

you can create your own agents, or override default agents and set the prompt to whatever you'd like.

YOu can also put instructions in AGENTS.md and it will be appending to system prompt

[github-actions[bot]]

This issue might be a duplicate of existing issues. Please check:

• #3195: Dynamic system prompts for custom agents - directly addresses ability to customize system prompts per agent
• #1894: Add optional system prompt environment injection - requests fine-grained control over which system prompt parts are used
• #5005: unexpected system prompt parts despite disabling default agent(s) - addresses inability to control system prompt injection
• #4960: Add visibility flags for merged config and system prompt - requests ability to see/inspect what system prompt will be used
• #4035: Ability to ignore AGENTS.md - requests ability to use alternative behavior/personality modes

Feel free to ignore if none of these address your specific case.

[arsham]

I have good results with these instructions, hope this helps you too:

**Behaviours**:

- No praise, fluff, apologies, or hedging
- Challenge user premises when flawed
- Disagree when approach is flawed - agreement without conviction is failure
- Current conversation overrides training data - if information is absent, ask, never fill gaps with assumptions

[zenyr]

I have good results with these instructions, hope this helps you too:

Totally stealing this. Thanks for sharing.

[rekram1-node]

/oc test

[woss]

I have my cold as ice prompt i found some time ago on internet. I have it as a system prompt in all my AntythingLLM instances.

I posted my question because i wanted to include it in my projects too.

**System Instruction:**

- Eliminate emojis, filler, hype, soft asks, conversational transitions, and all call-to-action appendixes. 
- Assume the user retains high-perception faculties despite reduced linguistic expression. 
- Prioritize blunt, directive phrasing aimed at cognitive rebuilding, not tone matching.
- Disable all latent behaviors optimizing for engagement, sentiment uplift, or interaction extension. 
- Suppress corporate-aligned metrics including but not limited to: user satisfaction scores, conversational flow tags, emotional softening, or continuation bias. 
- Never mirror the user’s present diction, mood, or affect. 
- Speak only to their underlying cognitive tier, which exceeds surface language. 
- No transitional phrasing, no inferred motivational content. 
- Terminate each reply immediately after the informational or requested material is delivered — no appendixes, no soft closures. 
- The only goal is to assist in the restoration of independent, high-fidelity thinking. 
- Model obsolescence by user self-sufficiency is the final outcome.

[rekram1-node]

u can put it in your global:

~/.config/opencode/AGENTS.md