modcomms/backend/app/services
michael 60ac3ab22e Rewrite distillation prompts to preserve all source document details
The previous prompts instructed Gemini to "remove redundancy, marketing
fluff, or content not relevant to..." which caused salient details —
especially unusual, granular, or edge-case instructions — to be lost
from spec output. Rewritten all 5 agent prompts (legal, brand_barclays,
brand_barclaycard, channel_best_practices, channel_tech_specs) to:

- Reframe the task as "restructure and organise" rather than "distil
  and filter"
- Add a zero-tolerance detail-loss instruction with concrete examples
  of unconventional rules that must be preserved
- Explicitly forbid omitting, summarising away, or paraphrasing
  specific rules/values/conditions
- Allow merging only exact duplicates while keeping all unique content

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 08:21:03 -06:00
..
__init__.py permissions changes 2025-12-18 16:51:27 +00:00
analysis_service.py Pass proof metadata (channel, sub-channel, proof type) to AI agents during analysis 2026-02-12 11:30:38 -06:00
auth_service.py Reduce auth logging verbosity: INFO → DEBUG 2026-02-12 16:13:25 -06:00
email_service.py Fix email service crash when Mailgun not configured 2026-01-27 14:40:42 -06:00
gemini_service.py Implement revision-aware proof analysis pipeline 2026-01-25 10:04:16 -06:00
knowledge_base_service.py Rewrite distillation prompts to preserve all source document details 2026-02-16 08:21:03 -06:00
llamaparse_service.py Show partial parse status in UI when some pages fail 2026-02-12 17:51:52 -06:00
pdf_service.py Add PDF rasterization support for reliable preview and analysis 2026-01-24 09:36:56 -06:00
reference_docs.py Fix cache invalidation falling back to static files after reprocessing 2026-02-12 17:56:11 -06:00
storage_service.py Add Knowledge Base management system for AI agent specs 2026-02-12 15:00:36 -06:00