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

82%

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

65%

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

The body is highly actionable with abundant executable examples, but it underuses its own reference file by duplicating index-type content inline and never signaling the reference, and the workflow lacks explicit validation checkpoints. Conciseness also slips via time-sensitive star counts and redundant sections.

Suggestions

Replace the inline 'Index types' section with a brief overview plus a clear pointer to references/index_types.md (e.g., 'See [index_types.md](references/index_types.md) for the full selection guide and tuning details') to fix the duplication and signal the bundle file.

Remove time-sensitive padding such as the '31,700+ GitHub stars' metric and consolidate the duplicated Performance tables so the body stays lean and durable.

Add an explicit validation step to the workflow — e.g., after search, verify recall by sweeping nprobe/efSearch values (as shown in the reference) before committing to an index choice — to reach the top workflow-clarity anchor.

Dimension	Reasoning	Score
Conciseness	The body is mostly lean executable code, but it duplicates the index-type material that already lives in references/index_types.md and includes time-sensitive fluff such as '31,700+ GitHub stars' and padding like '10-100× faster than CPU'; not score-1 verbose (it never explains what a vector or k-NN is) but not fully tight either.	2 / 3
Actionability	Provides copy-paste-ready executable code for Flat, IVF, HNSW, PQ, save/load, single/multi-GPU, and LangChain/LlamaIndex integration, matching the top anchor for fully executable concrete examples.	3 / 3
Workflow Clarity	Each index example sequences create → train → add → search and best-practice #1 gives a size-based selection heuristic, but there is no unified workflow with validation/checkpoint steps (e.g., tuning nprobe/efSearch to verify recall), so it sits at the middle anchor.	2 / 3
Progressive Disclosure	A bundle file references/index_types.md exists but is never linked from the body, while the inline 'Index types' section duplicates content that should live in that reference — matching the middle anchor of reference present but not signaled and content kept inline.	2 / 3
	Total	9 / 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, third-person, and explicitly states both capabilities and use-when triggers, distinguishing FAISS from metadata-aware vector databases. It satisfies the highest anchor on every dimension.

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', 'large-scale vector retrieval' — matching the top anchor for specific concrete actions.	3 / 3
Completeness	Explicitly answers both what ('efficient similarity search and clustering of dense vectors', supported index types) and when ('Use for fast k-NN search, large-scale vector retrieval, or when you need pure similarity search without metadata'), satisfying the top anchor with explicit triggers.	3 / 3
Trigger Term Quality	Covers natural terms a user would say — 'similarity search', 'k-NN search', 'vector retrieval', 'GPU acceleration', 'pure similarity search without metadata' — and uses third person ('Supports', 'Use for') with no first/second person, so the voice penalty does not apply.	3 / 3
Distinctiveness Conflict Risk	The 'pure similarity search without metadata' framing carves a clear niche and the body elsewhere contrasts with metadata-filtering stores (Chroma/Pinecone/Weaviate), making it unlikely to trigger for the wrong skill.	3 / 3
	Total	12 / 12 Passed

Validation

81%

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 — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	13 / 16 Passed

Repository: NousResearch/hermes-agent
Path: optional-skills/mlops/faiss/SKILL.md
Commit: 4c9628e

Reviewed: about 14 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.