Rag Improvement.task

GPTClaudeDeepSeek··▲ 525 copies·updated 2026-07-14

rag-improvement-task.prompt

# Task Prompt: RAG Improvement

## Objective

Improve retrieval and grounded generation quality for {{goal_dataset_or_domain}}.

## Required Inputs

- baseline metrics (at least recall@k and groundedness)
- corpus manifest
- target query set
- latency and cost bounds

## Execution Checklist

1. Measure baseline quality.
2. Diagnose likely bottlenecks: ingestion, chunking, retrieval, reranking, or generation prompt.
3. Apply one controlled change at a time where possible.
4. Re-evaluate against baseline.
5. Run safety checks for hallucination and unsupported citation risk.

## Completion Criteria

- measurable quality gain vs baseline
- no safety regressions
- no unacceptable latency or cost regression

fill the variables

This prompt has 1 variable. Pro fills them into a ready-to-paste prompt for you — no manual find-and-replace.

{{goal_dataset_or_domain}

Unlock with Pro →

when to use it

Community prompt sourced from the open-source GitHub repo Shuvam-Banerji-Seal/LLM-Whisperer (MIT). A "Rag Improvement.task" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.