rag-pipeline

Use this skill when building retrieval-augmented generation (RAG) systems — document ingestion, chunking, embedding, vector search, and grounded responses.

Quality

76%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./external/anthropic-official/rag-pipeline/SKILL.md

Quality

Content

68%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-structured, code-first RAG skill that is lean and actionable, with clear sequencing of the pipeline stages. Its main gaps are a stale/invalid model id that breaks copy-paste runnability, missing validation for the batch indexing step, and all content inlined with no progressive disclosure to reference files.

Suggestions

Replace the invalid model id 'claude-opus-4-6' with a current Claude model id (e.g. 'claude-opus-4-8') so the query code runs as written.

Add a validation checkpoint in the batch indexing path — e.g. assert embeddings are non-empty and chunk count matches before collection.add, and verify a sample query retrieves results — to lift workflow clarity above the batch-cap of 3.

Move the unsupported vector-DB and embedding-model variants (Pinecone/Weaviate, other embedding providers) referenced in the Quick Reference table into a separate reference file, keeping SKILL.md a lean overview with one-level-deep links.

Dimension	Reasoning	Score
Conciseness	The body is lean and mostly code with no padding of concepts Claude already knows, but it inlines a stale model id ('claude-opus-4-6') without a deprecated/old-patterns section and the Quick Start partially re-demonstrates earlier code, keeping it just below a 5.	4 / 5
Actionability	Provides concrete, copy-paste-ready functions for the full pipeline plus a runnable Quick Start and reference table, but rag_query uses an invalid model id ('claude-opus-4-6') that would error at runtime and the table advertises DBs/embedding models the code never implements, so it is not fully executable as written.	4 / 5
Workflow Clarity	The ingest→chunk→embed→index→retrieve→generate sequence is clearly laid out, but index_document is a batch operation with no validation checkpoints (no embedding/count verification, no failure handling), and the rubric caps batch workflows without validation at 3.	3 / 5
Progressive Disclosure	Sections are well-organized with clear headers and no nested references, but at ~110 lines with all content inline and no external reference files, it exceeds the simple-skill threshold and could split out the unsupported vector-DB/embedding-model options hinted at in the table.	4 / 5
	Total	15 / 20 Passed

Description

83%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, specific description that clearly scopes a RAG-building skill with an explicit 'Use when' trigger and low conflict risk. Its main weakness is that the 'when' clause restates the domain rather than enumerating concrete user-mention triggers, and a few natural synonyms are missing.

Suggestions

Expand the 'when' clause with concrete user-mention triggers, e.g. 'Use when the user asks to chat with their documents, search a document corpus, or build a question-answering system over their data.'

Add natural synonyms and phrasings users actually say (e.g. 'chat with my documents', 'semantic search', 'Q&A over my data') alongside the technical terms.

Dimension	Reasoning	Score
Specificity	Lists five concrete actions — 'document ingestion, chunking, embedding, vector search, and grounded responses' — giving comprehensive coverage of the RAG pipeline rather than vague domain naming.	5 / 5
Completeness	Both 'what' (the pipeline stages) and 'when' ('Use this skill when building ... RAG systems') are present, but the 'when' restates the domain rather than offering concrete user-mention triggers, so it is not a 5.	4 / 5
Trigger Term Quality	Includes natural domain terms ('RAG', 'chunking', 'embedding', 'vector search') users would say, but misses common synonyms and user phrasings like 'chat with my documents' or 'search my docs', so it falls short of comprehensive.	4 / 5
Distinctiveness Conflict Risk	'Retrieval-augmented generation (RAG) systems' is a clear, well-recognized niche with distinct triggers (embeddings, vector search), giving minimal overlap risk with unrelated skills.	5 / 5
	Total	18 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: ProwlrBot/prowlr-marketplace
Path: external/anthropic-official/rag-pipeline/SKILL.md
Commit: 32f0050

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.