Oliver/modcomms
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Make Gemini analysis responses concise and actionable

Browse source 
Add explicit formatting instructions to agent prompts requesting bullet-point
output instead of verbose paragraphs. Update JSON schema descriptions for
feedback and summary fields to enforce concise, outline-style format.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
michael 2026-01-24 09:48:08 -06:00
parent e094014dbe
commit ec2fb82205
4 changed files with 23 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
backend/app/agents/brand_agent.py
View file

@ -62,6 +62,13 @@ RAG Status Guidelines:
- **Red**: Significant brand guideline violations that must be fixed before use
If the proof is nonsensical, not a marketing material, or cannot be analyzed, set analysisStatus to 'low_confidence'.
**Response Format:**
- Keep feedback brief and scannable
- Use bullet points for each finding
- Each bullet should be one actionable sentence
- Start with the issue, then the recommendation
- Example: "Logo placement incorrect (bottom-left) - move to top-right corner per guidelines"
"""
# Use single-image or multi-image analysis depending on input

7
backend/app/agents/channel_agent.py
View file

@ -80,6 +80,13 @@ RAG Status Guidelines:
- **Red**: Significant technical issues that will impact display or accessibility
If the proof is nonsensical, not a marketing material, or cannot be analyzed, set analysisStatus to 'low_confidence'.
**Response Format:**
- Keep feedback brief and scannable
- Use bullet points for each finding
- Each bullet should be one actionable sentence
- Start with the issue, then the recommendation
- Example: "Logo placement incorrect (bottom-left) - move to top-right corner per guidelines"
"""
# Use single-image or multi-image analysis depending on input

7
backend/app/agents/lead_agent.py
View file

@ -68,7 +68,12 @@ Your summary should:
- For an 'Analysis Error' verdict, explain that the proof could not be reliably processed and has been logged for human review. Advise the user to try again with a revised proof.
- For a 'Failed' status, highlight the critical 'Red' issues that must be addressed.
- For a 'Passed' status, mention any 'Amber' areas for consideration, if they exist, while maintaining an encouraging tone.
- Be professional, clear, and constructive.
**Response Format:**
- Start with a one-line verdict statement
- List key issues as bullet points (max 3-5 bullets)
- Each bullet: one sentence, actionable
- For 'Passed': briefly note any amber items in 1-2 bullets
Here are the specialist reviews:
{self._format_reviews(reviews)}

6
backend/app/services/gemini_service.py
View file

@ -66,7 +66,7 @@ class GeminiService:
},
"feedback": {
"type": "string",
"description": "Constructive, professional feedback explaining the RAG status and highlighting both positive aspects and areas for improvement."
"description": "Brief bullet-point feedback. Each bullet: one actionable sentence. Format: '• Issue: recommendation'. Max 5 bullets."
},
"issues": {
"type": "array",
@ -167,7 +167,7 @@ class GeminiService:
},
"feedback": {
"type": "string",
"description": "Constructive, professional feedback explaining the RAG status and highlighting both positive aspects and areas for improvement."
"description": "Brief bullet-point feedback. Each bullet: one actionable sentence. Format: '• Issue: recommendation'. Max 5 bullets."
},
"issues": {
"type": "array",
@ -252,7 +252,7 @@ class GeminiService:
},
"summary": {
"type": "string",
"description": "A concise, professional summary explaining the overall status, based on the specialist reviews."
"description": "Brief summary: one-line verdict, then 3-5 bullet points listing key issues/recommendations."
}
},
"required": ["overallStatus", "summary"]