rag-architect

Designs and implements production-grade RAG systems by chunking documents, generating embeddings, configuring vector stores, building hybrid search pipelines, applying reranking, and evaluating retrieval quality. Use when building RAG systems, vector databases, or knowledge-grounded AI applications requiring semantic search, document retrieval, context augmentation, similarity search, or embedding-based indexing.

Quality

88%

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Advisory

Suggest reviewing before use

Quality

Content

77%

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

This is a well-structured RAG skill with strong actionability — the executable code examples covering the full RAG pipeline (chunking, embedding, hybrid search, reranking, evaluation) are its greatest strength. The workflow is clear with explicit validation checkpoints at each stage. The main weaknesses are moderate verbosity in the constraints/output sections and the absence of the referenced bundle files, which undermines the progressive disclosure structure.

Suggestions

Trim the MUST DO/MUST NOT DO lists to only non-obvious, domain-specific constraints (e.g., remove 'handle edge cases' and 'monitor latency' which Claude already knows).

Provide the 5 referenced files (vector-databases.md, embedding-models.md, etc.) as bundle files, and move some of the inline code examples into those references to keep SKILL.md as a concise overview.

Dimension	Reasoning	Score
Conciseness	The skill is mostly efficient with executable code examples, but includes some unnecessary verbosity in the constraints section (listing 8 MUST DO and 8 MUST NOT DO items, some of which are obvious to Claude like 'handle edge cases') and the output templates section adds moderate bloat. The code comments are generally useful but a few are redundant.	2 / 3
Actionability	Provides fully executable, copy-paste ready Python code for all major RAG components: chunking with LangChain, embedding with OpenAI, indexing with Qdrant, hybrid search with BM25+RRF, reranking with Cohere, and evaluation with RAGAS. Each example includes concrete imports, function signatures, and realistic parameters.	3 / 3
Workflow Clarity	The 5-step core workflow is clearly sequenced with explicit validation checkpoints after each implementation step (assert statements for metadata, deduplication, search results, and metric thresholds). The instruction 'validate before moving on' with concrete checkpoint assertions creates proper feedback loops for error recovery.	3 / 3
Progressive Disclosure	The reference table clearly signals 5 separate reference files with 'Load When' context, which is excellent structure. However, no bundle files are provided, meaning all those references are broken/non-existent. The main SKILL.md also includes substantial inline code that could arguably be in the reference files, making the body longer than necessary for an overview.	2 / 3
	Total	10 / 12 Passed

Description

100%

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

This is an excellent skill description that clearly defines a specific domain (RAG systems), lists concrete capabilities, and provides explicit trigger guidance via a well-constructed 'Use when' clause. The description uses proper third-person voice and includes a comprehensive set of natural trigger terms that practitioners would use. It serves as a strong example of a well-crafted skill description.

Dimension	Reasoning	Score
Specificity	Lists multiple specific concrete actions: chunking documents, generating embeddings, configuring vector stores, building hybrid search pipelines, applying reranking, and evaluating retrieval quality. These are all distinct, actionable capabilities.	3 / 3
Completeness	Clearly answers both 'what' (designs and implements production-grade RAG systems with specific sub-tasks listed) and 'when' (explicit 'Use when' clause covering RAG systems, vector databases, knowledge-grounded AI applications, and multiple trigger scenarios).	3 / 3
Trigger Term Quality	Excellent coverage of natural terms users would say: 'RAG systems', 'vector databases', 'semantic search', 'document retrieval', 'embeddings', 'similarity search', 'embedding-based indexing', 'context augmentation'. These are terms practitioners naturally use when seeking help with this domain.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear niche around RAG/retrieval-augmented generation systems with highly specific triggers like 'vector databases', 'reranking', 'hybrid search pipelines', and 'embedding-based indexing' that are unlikely to conflict with general coding or document processing skills.	3 / 3
	Total	12 / 12 Passed

Validation

100%

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 — 11 / 11 Passed

Validation for skill structure

No warnings or errors.

Repository: Jeffallan/claude-skills
Commit: e8be415

Reviewed: 4 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.