mamba-architecture

State-space model with O(n) complexity vs Transformers' O(n²). 5× faster inference, million-token sequences, no KV cache. Selective SSM with hardware-aware design. Mamba-1 (d_state=16) and Mamba-2 (d_state=128, multi-head). Models 130M-2.8B on HuggingFace.

Quality

64%

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 ./skills/model-architecture/mamba/SKILL.md

Quality

Content

60%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.

The body is well-structured and highly actionable with executable examples and useful troubleshooting, but it suffers from repetition that hurts conciseness and, more seriously, references three files that do not exist in the bundle, breaking progressive-disclosure navigation.

Suggestions

Fix the Advanced topics references to point to the real bundle files (architecture-details.md, benchmarks.md, training-guide.md), or add the missing selective-ssm.md, mamba2-details.md, and performance.md files.

De-duplicate the '5× faster inference' and 'no KV cache' claims — state them once and reference that section instead of repeating in Quick start, Workflow 4, Hardware requirements, and Performance.

Add an explicit validation step to the benchmark and generation workflows (e.g., assert output shape, verify model loads before generate) to introduce clear feedback loops.

Dimension	Reasoning	Score
Conciseness	Mostly efficient with executable code and no over-explanation of basic concepts, but '5× faster inference' and 'no KV cache' are repeated across Quick start, Workflow 4, Hardware requirements, and Performance sections, and the Advantages/Performance lists add padding that could be trimmed.	3 / 5
Actionability	Provides mostly executable, copy-paste-ready code and commands across the Mamba block, LM config+generate, pretrained loading, Mamba-1 vs Mamba-2 instantiation, and troubleshooting, with minor gaps such as the Transformer benchmark reusing a Mamba script path and generate() args varying by library version.	4 / 5
Workflow Clarity	Workflows are clearly numbered and sequenced (Workflow 1-4), and the Common issues section supplies error-recovery guidance for CUDA OOM and missing causal-conv1d; however there are no explicit validate-then-proceed checkpoints, which keeps it just below 5.	4 / 5
Progressive Disclosure	An 'Advanced topics' section cleanly signals one-level-deep references to selective-ssm.md, mamba2-details.md, and performance.md, but none of those files exist — the actual bundle contains architecture-details.md, benchmarks.md, and training-guide.md — so the navigation links are broken.	2 / 5
	Total	13 / 20 Passed

Description

70%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 specific, distinctive, and rich in natural trigger terms, but it omits any explicit 'Use when...' guidance telling Claude when to invoke the skill, which caps completeness. Adding a trigger clause would raise the completeness and overall score.

Suggestions

Append an explicit 'Use when...' clause (e.g., 'Use when working with Mamba/state-space models, linear-complexity sequence modeling, long-context inference, or comparing SSMs to Transformers.') so Claude knows when to trigger this skill.

Frame capabilities as concrete actions the skill performs (e.g., 'Builds, loads, benchmarks, and compares Mamba SSMs') rather than only listing architecture properties.

Add common synonyms/file extensions users might say (e.g., '.bin checkpoints', 'Mamba-2 checkpoints') to broaden trigger-term coverage.

Dimension	Reasoning	Score
Specificity	Lists many concrete, specific claims ('O(n) complexity vs Transformers' O(n²)', '5× faster inference', 'million-token sequences', 'no KV cache', 'Mamba-1 (d_state=16) and Mamba-2 (d_state=128, multi-head)', 'Models 130M-2.8B on HuggingFace') with comprehensive coverage, though these are architecture properties rather than discrete user-facing actions.	4 / 5
Completeness	Clearly answers 'what' (state-space model, O(n) complexity, Mamba-1/Mamba-2 variants, available model sizes) but has no 'Use when...' clause or equivalent explicit trigger guidance, capping completeness at 3 per the rubric.	3 / 5
Trigger Term Quality	Strong natural keyword coverage ('Mamba', 'state-space model', 'SSM', 'Transformers', 'KV cache', 'HuggingFace', 'linear complexity', 'long context') that users would actually say, but missing common synonyms and file extensions.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (selective state-space models vs Transformers) with distinct triggers and minimal conflict risk with other skills.	5 / 5
	Total	16 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 3 missing	Warning
referenced_paths_exist	Referenced path issues: 6 missing	Warning

	Total	13 / 16 Passed

Repository: OpenLAIR/dr-claw
Path: skills/model-architecture/mamba/SKILL.md
Commit: 4b35756

Reviewed: 1 day 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.