faiss

Fast vector similarity search at billion scale.

Quality

55%

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 ./optional-skills/mlops/faiss/SKILL.md

Quality

Content

65%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 strong, code-dense quick-reference: executable examples for every major operation and useful comparison tables, with only light marketing padding. Its two real weaknesses are the missing explicit decision workflow (which index type for which use case exists only as a one-liner) and an orphaned references/index_types.md whose content is duplicated inline in SKILL.md.

Suggestions

Replace the inlined 'Index types' section with a one-line decision table (dataset size → index type) plus a link to references/index_types.md, e.g. '**Index guide**: See [index_types.md](references/index_types.md) for all four index types with full code, training requirements, and tradeoffs.'

Add an explicit index-selection workflow with a checkpoint, e.g. '1. Match dataset size to index type (table below) → 2. If IVF/PQ: train before add (train() raises on insufficient data) → 3. Verify recall on a held-out query set before deploying.'

Fix incomplete examples: define 'docs' in the LangChain snippet (e.g. text_splitter output) and add a search call to the PQ example so every code block is fully copy-paste runnable; trim the 'Metrics' star counts and Resources star repetition.

Dimension	Reasoning	Score
Conciseness	The body is code-forward and lean — no explanations of concepts Claude already knows — with only minor padding: the 'Metrics' star counts ('31,700+ GitHub stars'), the Resources section repeating them, and marketing claims like '10-100× faster'. This fits 'efficient; minor instances of over-explanation that could be trimmed' rather than score 5, which would require dropping those sections.	4 / 5
Actionability	Nearly all guidance is executable, copy-paste-ready code (install, basic search, four index types, save/load, GPU transfer, LangChain/LlamaIndex). Minor gaps keep it from score 5: the LangChain example uses undefined variables ('docs'), the PQ example lacks a search step, and the LlamaIndex snippet is a fragment.	4 / 5
Workflow Clarity	The document is a reference catalog rather than a sequenced workflow: the train→add→set-nprobe→search sequence for IVF/PQ is only implicit in code, and index-type selection guidance is a one-line best-practice bullet rather than an explicit decision flow. There are no validation checkpoints, though no destructive/batch operations demand them — matching 'sequence present but checkpoints missing or implicit'.	3 / 5
Progressive Disclosure	The bundle provides references/index_types.md, but the body inlines ~55 lines of index-type detail (Flat/IVF/HNSW/PQ) and never links to that file, leaving the reference orphaned and its content duplicated. This matches 'content that should be separate is inline' with references present in the bundle but not signaled — not score 4, since navigation to the existing reference file is impossible.	3 / 5
	Total	14 / 20 Passed

Description

45%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.

The description is concise and names a specific niche, but it reads more like a tagline than a skill trigger: it has no concrete capability listing and no 'Use when' trigger clause. Users asking for 'nearest neighbor search' or 'embedding search' would only partially match.

Suggestions

List 2-3 concrete actions in the description, e.g. 'Build, train, and search FAISS indexes (Flat, IVF, HNSW, PQ); run on GPU; persist indexes to disk.'

Add an explicit trigger clause, e.g. 'Use when the user mentions FAISS, vector similarity search, nearest-neighbor (k-NN) search, or embedding retrieval at large scale.'

Include natural synonyms users actually say — 'nearest neighbor', 'k-NN', 'embeddings', 'vector database' — to improve trigger matching beyond the single phrase 'vector similarity search'.

Dimension	Reasoning	Score
Specificity	The description names the domain ('Fast vector similarity search') and scale ('at billion scale') but lists no concrete actions such as building indexes, searching, or GPU offload. It matches the anchor 'names the domain but actions are minimal or generic' — not score 1 (not pure abstraction) and not score 3 (no explicit 1-2 concrete actions are stated).	2 / 5
Completeness	The 'what' is clear (fast billion-scale vector similarity search) but there is no 'Use when...' clause or equivalent trigger guidance, which caps completeness at 3 per the judging guidelines. It is not score 2 because the 'what' is specific, not vague.	3 / 5
Trigger Term Quality	'vector similarity search' is a natural phrase users would say, but the description misses common variations and synonyms users would actually use: 'nearest neighbor', 'k-NN', 'embeddings', 'FAISS', 'vector database'. This fits 'some relevant keywords but missing common variations or synonyms' rather than score 4's 'good keyword coverage'.	3 / 5
Distinctiveness Conflict Risk	The vector-similarity-search niche is somewhat specific, but the description would also plausibly match other vector-store skills (Chroma, Pinecone, Annoy, Weaviate) — the skill body itself lists these as alternatives. Anchor 3 ('somewhat specific but could still overlap with similar skills') is the best fit.	3 / 5
	Total	11 / 20 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: 68fa7e9

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.