faiss

Facebook's library for efficient similarity search and clustering of dense vectors. Supports billions of vectors, GPU acceleration, and various index types (Flat, IVF, HNSW). Use for fast k-NN search, large-scale vector retrieval, or when you need pure similarity search without metadata. Best for high-performance applications.

Quality

88%

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

Quality

Content

77%

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

The content is highly actionable with copy-paste-ready code, but it is padded with marketing-style metrics and basic definitions Claude already knows, and it fails to leverage its existing reference bundle — index-type material is duplicated inline instead of pointing to references/index_types.md. The main improvements are tightening promotional fluff and wiring up the reference file.

Suggestions

Link the existing references/index_types.md from the body (e.g., under 'Index types' add 'See [index_types.md](references/index_types.md) for the complete selection guide') and move the duplicated index detail there rather than keeping it inline.

Trim unneeded promotional/definitional content — GitHub star counts, '10-100× faster' claims, and glossary labels like 'HNSW (Hierarchical NSW)' — to raise conciseness.

Reconsider the Performance table: it largely repeats the per-index guidance already given; either consolidate it into the best-practices list or move it into the reference.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient with executable snippets, but pads in unnecessary promotion Claude does not need ('31,700+ GitHub stars', '10-100× faster', the performance table, and definitional labels like 'HNSW (Hierarchical NSW)'); not the lean level-3 standard. Not level 1 because the code blocks genuinely earn their place.	2 / 3
Actionability	It provides complete, copy-paste-ready executable code for every major path — IndexFlatL2/IVF/HNSW/PQ, save/load, GPU transfer, and LangChain/LlamaIndex integration. Not below 3 because nothing is pseudocode or abstract.	3 / 3
Workflow Clarity	For a reference/library skill each single action (create index, add, search, save/load) is unambiguous, satisfying the simple-skill allowance; there is no destructive multi-step operation requiring validation checkpoints. Not below 3 because the rubric permits a 3 for single-purpose skills whose actions are clear.	3 / 3
Progressive Disclosure	A reference bundle (references/index_types.md, a 280-line index-types guide) exists but is never linked or signaled from the body, while the body duplicates index-type content inline — content that should be separate is inline and the reference is not clearly signaled. Not level 1 because the body is reasonably sectioned and not a monolithic wall; not level 3 because navigation to the reference is absent.	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.

The description is specific, uses natural trigger terms, explicitly answers both what and when, and is clearly distinct from richer vector-database skills. Voice is third person with no first/second-person slip. It does not need improvement on the description dimensions.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capabilities: 'similarity search and clustering of dense vectors', 'GPU acceleration', 'various index types (Flat, IVF, HNSW)', 'fast k-NN search', and 'large-scale vector retrieval'.	3 / 3
Completeness	It answers both 'what' (similarity search/clustering of dense vectors with listed index types) and 'when' via the explicit 'Use for ... or when you need pure similarity search without metadata' trigger clause.	3 / 3
Trigger Term Quality	Natural terms a user would say are well covered — 'fast k-NN search', 'vector retrieval', 'similarity search', 'clustering', 'GPU acceleration' — rather than jargon-only phrasing.	3 / 3
Distinctiveness Conflict Risk	It carves a clear niche — 'pure similarity search without metadata' — which distinguishes it from full vector databases that add metadata filtering, making wrong-skill triggering unlikely.	3 / 3
	Total	12 / 12 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: Orchestra-Research/AI-Research-SKILLs
Path: 15-rag/faiss/SKILL.md
Commit: 773a529

Reviewed: about 12 hours 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.