Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add JayRHa/AgentSkills --skill rag-pipeline-designergit clone --depth 1 https://github.com/JayRHa/AgentSkillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/jayrha/agentskills/rag-pipeline-designer)<a href="https://agentmods.dev/skills/jayrha/agentskills/rag-pipeline-designer"><img src="https://agentmods.dev/badge/skills/jayrha/agentskills/rag-pipeline-designer.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00121 | $0.02449 |
| Opus 5 | $0.00060 | $0.01224 |
| Sonnet 5 | $0.00024 | $0.00490 |
| Haiku 4.5 | $0.00012 | $0.00245 |
Grade A, and why
rag-pipeline-designer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.
RAG Pipeline Designer
Overview
Keywords: RAG, retrieval-augmented generation, chunking, embeddings, vector database, hybrid search, BM25, dense retrieval, reranking, cross-encoder, recall@k, MRR, nDCG, faithfulness, context window, query expansion, semantic search.
This skill turns a vague "we want to do RAG over our docs" into a concrete, defensible architecture with tradeoffs made explicit. A RAG pipeline is a sequence of lossy stages — each stage can silently throw away the right answer. The job is to identify which stage is failing (or will fail) and pick components that fit the corpus, query distribution, latency budget, and cost ceiling.
Treat RAG design as five decisions made in order, with evaluation wrapped around all of them:
- Ingestion & chunking — how documents become retrievable units.
- Embedding & indexing — how units become searchable vectors (and/or keyword index).
- Retrieval — how a query fetches candidates.
- Reranking & context assembly — how candidates are ordered and packed into the prompt.
- Generation — how the LLM is grounded and constrained.
- Evaluation — how you prove each stage works, offline, before shipping.
Read references/component-catalog.md for concrete component options and references/evaluation-metrics.md for the metric definitions. Use scripts/chunk_text.py to produce candidate chunkings and scripts/eval_retrieval.py to score retrieval against a labeled set. Fill in templates/rag-design-doc.md to capture the final design. See examples/support-kb-rag.md for a complete worked design.
Workflow
Follow these steps. Do not skip evaluation — a RAG system without an eval set is undebuggable.
- Characterize the corpus and queries. Before choosing anything, answer: How many documents and total tokens? What format (PDF, HTML, Markdown, code, tables, transcripts)? How long is a typical document? Are answers usually in one place or scattered? What do real user queries look like — keyword lookups, natural questions, multi-hop? What is the latency budget (interactive <1s vs batch) and cost ceiling? Capture answers in the design doc's "Context" section.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 124 lines · 121 tokens per session scan A 4fd7d32e67ae
rag-pipeline-designer is a skill published in the GitHub repository JayRHa/AgentSkills (5 stars, last pushed 1mo ago), licensed MIT. It adds 121 tokens to every session and 2,449 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
rag-architect
Use when building RAG systems, vector databases, or knowledge-grounded AI applications requiring semantic search, document retrieval, or context augmentation.
dspy
DSPy: declarative LM programs, auto-optimize prompts, RAG.
qdrant
Vector search engine for production RAG systems.
chroma
Embedding database for RAG and semantic search.
pinecone
Managed vector DB for production RAG and search.
faiss
Fast vector similarity search at billion scale.