long-context

Extend context windows of transformer models using RoPE, YaRN, ALiBi, and position interpolation techniques. Use when processing long documents (32k-128k+ tokens), extending pre-trained models beyond original context limits, or implementing efficient positional encodings. Covers rotary embeddings, attention biases, interpolation methods, and extrapolation strategies for LLMs.

Quality

74%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./backend/cli/skills/llm-tools/long-context/SKILL.md

Quality

Content

61%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 content is a well-structured, code-heavy reference with good progressive disclosure to real bundle files. Its main weaknesses are restating concepts Claude already knows in 'Core Concepts' and the absence of validation checkpoints in the fine-tuning/deployment workflows.

Suggestions

Trim or relocate the 'Core Concepts' explanations and math formulations to references/rope.md and references/extension_methods.md, since Claude already knows RoPE/ALiBi properties; keep only what is non-obvious.

Add explicit validation checkpoints to the fine-tuning workflow (e.g., evaluate perplexity on a held-out long-context set and only proceed/serve if it stays below a threshold), since fine-tuning is a batch operation.

Resolve undefined variables in code examples (query, key, attn_scores, F, long_document_dataset) or label them as placeholders so the snippets are fully copy-paste ready.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient with substantial executable code, but the 'Core Concepts' section restates properties and math formulations (e.g., RoPE/ALiBi advantages, derivation formulas) that Claude already knows, which the rubric explicitly penalizes.	3 / 5
Actionability	Provides substantial copy-paste-ready executable code across RoPE, ALiBi, position interpolation, fine-tuning, and deployment, with only minor gaps (undefined vars like query, attn_scores, F, long_document_dataset).	4 / 5
Workflow Clarity	Steps are roughly sequenced (install → quick start → fine-tune → deploy), but fine-tuning is a batch operation with no explicit validation/verification checkpoints (e.g., perplexity check on long sequences), which caps this dimension at 3 per the rubric.	3 / 5
Progressive Disclosure	Clear sectioned structure with real, well-signaled one-level reference files (references/rope.md, extension_methods.md, fine_tuning.md, all verified to exist), though some inline 'Core Concepts' material could arguably live in the reference files.	4 / 5
	Total	14 / 20 Passed

Description

87%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 strong: it names the domain, multiple concrete techniques, and provides an explicit 'Use when' clause with concrete triggers in third-person voice. Minor room for improvement lies in adding a few more natural synonyms and slightly more concrete actions.

Dimension	Reasoning	Score
Specificity	Lists several specific techniques ("RoPE, YaRN, ALiBi, and position interpolation") and concrete actions (extend, process, implement), but the actions are somewhat more abstract than the rubric's file-operation examples, so it sits below the 5 anchor.	4 / 5
Completeness	Clearly answers 'what' ("Extend context windows of transformer models using RoPE, YaRN, ALiBi, and position interpolation techniques") and 'when' ("Use when processing long documents (32k-128k+ tokens), extending pre-trained models beyond original context limits, or implementing efficient positional encodings") with concrete trigger phrases.	5 / 5
Trigger Term Quality	Good natural keyword coverage ("long documents", "context windows", "32k-128k+ tokens", "positional encodings", "extending pre-trained models") that a practitioner would say, but a few common synonyms are missing, keeping it below the comprehensive 5 anchor.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (transformer long-context extension) with distinct technique-named triggers and minimal overlap risk with other skills.	5 / 5
	Total	18 / 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
skill_md_line_count	SKILL.md is long (538 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
referenced_paths_exist	Referenced path issues: 1 missing	Warning

	Total	13 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/llm-tools/long-context/SKILL.md
Commit: fcd2e11

Reviewed: 2 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.