More screenshots
Works with the AI Agents you already use
RAG Essentials
Build a complete RAG pipeline, from indexing to evaluation. 10 ordered skills, starting with Retrieval-Augmented Generation (RAG), Vector Databases, Embedding Engineering.
$3
RAG Essentials
Example session with this skill installed
My RAG system is hallucinating. The documents are there, but the model gives wrong answers. How do I fix the retrieval and measure the improvement?
- Read your context and instructions
- Compiled the rag essentials
I will diagnose your RAG pipeline by separating retrieval quality from generation quality and setting up a measurement protocol.
RAG Diagnosis and Evaluation Plan
| Stage | Metric | Goal |
|---|---|---|
| Retrieval | Recall@5 | Ensure the correct chunk is in the top 5 results |
| Groundedness | Faithfulness | Ensure the answer only uses provided context |
| Relevance | Answer Relevance | Ensure the answer actually addresses the user's query |
Recommended Technical Strategy
Implement Hybrid Search: We will combine BM25 for exact keyword matching (Product IDs, names) with vector search for semantic meaning.
2.
Apply Contextual Retrieval: We will use a "Contextual-BM25" approach where each chunk is prepended with a short document summary to prevent loss of context.
3.
Add a Reranker: We will use a cross-encoder (e.g., Cohere Rerank) to reorder the top 20 candidates before sending the top 5 to the LLM.
Measurement Protocol
We will build a "Golden Set" of 30 questions with known evidence spans. We will then run a comparison:
- Baseline: Current vector-only retrieval.
- Target: Hybrid search + Reranking.
- Success Criteria: A 15% increase in Recall@5 and a reduction in "context-lost" failure labels.
Next steps
- Extract 30 sample question-answer-evidence triplets from your logs.
- Run the
retrieval-testing-and-tuning-recall-kprotocol to establish your baseline recall. - Select a reranker model based on your latency budget (200ms vs 500ms).
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
People building assistants over their own documents often need a repeatable process, not a one-off prompt. This pack targets: Build a complete RAG pipeline, from indexing to evaluation.
What's inside (10 skills, suggested learning order)
- Retrieval-Augmented Generation (RAG) — Combine search over your own data with answer generation.
- Vector Databases — Store and query embeddings at scale.
- Embedding Engineering — Choose and tune embedding models.
- Chunking Strategies — Split documents for better retrieval.
- Hybrid Search (Vector + Lexical) — Combine BM25 and vectors for higher precision.
- Search Result Reranking — Reorder retrieved passages with cross-encoder models.
- Query Rewriting and Expansion — Rephrase questions for better retrieval.
- Contextual Retrieval (Per-Chunk Context) — Enrich each chunk with context to improve retrieval.
- RAG Pipeline Evaluation — Measure faithfulness, relevance and recall.
- Retrieval Testing and Tuning (recall@k) — Measure and tune retrieval quality with metrics.
Why a skill pack instead of a one-off prompt
Each skill gives the agent a defined trigger, step-by-step procedure and output format, so results are consistent across sessions and easier to review.
How to use
- Install the pack in your SKILL.md-compatible agent.
- Ask for the task in plain language (for example: "combine search over your own data with answer generation").
- Review the output against the skill's quality checks before using it.
Known limitations
- Outputs are drafts that need human review; they are not professional, legal or financial advice.
- Tools, APIs and prices change; check current vendor documentation before relying on version-specific details.
- Quality depends on the context you give the agent (data, requirements, constraints).
- Tested only where stated in the Compatibility note.
Also by Rodrigo Santos
Specialized Agents and Distribution; AI Agents: Fundamentals and Protocols; Model Fine-Tuning and Alignment
Does not support fine-tuning embedding models. Not for datasets that fit in a single context window. Requires a labeled query set for best results.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 4 days ago