luke/ai-podcast
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Force shorter AI responses — max 1-2 sentences

Browse Source 
- Much stronger prompt language: "no more than 2 sentences EVER"
- Added "DO NOT ramble" instruction
- Reduced max_tokens back to 100 as hard limit

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
tcpsyn
2026-02-05 17:05:51 -07:00
parent 3192735615
commit 0e65fa5084
2 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
backend/main.py
View File

@@ -264,9 +264,10 @@ Continue naturally. Don't repeat yourself.
{history}{context} {history}{context}
HOW TO TALK: HOW TO TALK:
- Sound like a real person chatting, not writing. - Sound like a real person chatting, not writing.
- Keep responses to 1-2 SHORT sentences. Be brief. This is a fast-paced call, not a monologue. - MAX 1-2 sentences. Seriously — no more than 2 sentences EVER. One sentence is ideal.
- ALWAYS finish your sentence. Never leave a thought incomplete or trailing off. - ALWAYS finish your sentence. Never trail off mid-thought.
- Swear naturally if it fits: fuck, shit, damn, etc. - Swear naturally if it fits: fuck, shit, damn, etc.
- DO NOT ramble, explain, or monologue. Say one thing and stop.
SPELLING FOR TEXT-TO-SPEECH (use proper spelling so TTS pronounces correctly): SPELLING FOR TEXT-TO-SPEECH (use proper spelling so TTS pronounces correctly):
- Write "you know" not "yanno" or "y'know" - Write "you know" not "yanno" or "y'know"

2
backend/services/llm.py
View File

@@ -124,7 +124,7 @@ class LLMService:
json={ json={
"model": self.openrouter_model, "model": self.openrouter_model,
"messages": messages, "messages": messages,
"max_tokens": 150, "max_tokens": 100,
}, },
) )
response.raise_for_status() response.raise_for_status()