regex-vs-llm-structured-text

在解析结构化文本时，用于在正则表达式（Regex）和大型语言模型（LLM）之间进行选择的决策框架——优先使用正则表达式，仅针对低置信度的边界情况引入 LLM。

Quality

72%

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 ./skills/regex-vs-llm-structured-text/SKILL.md

Quality

Content

87%

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

The body is a tight, actionable framework with executable code and clean organization, scoring high on conciseness, actionability, and progressive disclosure. Its only real gap is the absence of an explicit validation/retry feedback loop for the batch document-processing pipeline.

Suggestions

Add an explicit validation/retry step in process_document: after validate_with_llm, re-score confidence on corrected items and retry or flag items that still fall below threshold.

Complete validate_with_llm by parsing the LLM response (e.g., JSON or 'CORRECT') and returning a real ParsedItem instead of the undefined 'corrected_item' stub.

Optionally factor the long code blocks into a scripts/ bundle with a one-level reference from SKILL.md to keep the overview leaner, though the current single-file structure is already acceptable.

Dimension	Reasoning	Score
Conciseness	Lean and high-signal throughout: a one-line core insight, a compact decision tree, an architecture diagram, and tight code — no padding with concepts Claude already knows.	3 / 3
Actionability	Provides fully executable Python (parser, confidence scorer, validator, full pipeline) with real imports and dataclasses; the single 'corrected_item' stub in validate_with_llm is a minor gap that does not undermine the copy-paste-ready core.	3 / 3
Workflow Clarity	The pipeline is clearly sequenced with an explicit confidence checkpoint (≥0.95 vs <0.95), but this batch document operation lacks an explicit validate-output/fix/retry feedback loop on LLM-corrected results, capping the score at 2 per the batch-operations guideline.	2 / 3
Progressive Disclosure	Self-contained SKILL.md with no bundle files and no nested/deep references, organized into clear, well-labeled sections — appropriate single-file structure that needs no external references.	3 / 3
	Total	11 / 12 Passed

Description

57%

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 concrete and distinctive with a clear niche, but it states only the 'what' without an explicit 'Use when...' trigger and offers limited natural trigger-term variations. Adding an explicit when-to-use clause and broader keyword coverage would lift completeness and trigger quality.

Suggestions

Add an explicit 'Use when...' clause (e.g., 'Use when parsing structured/repetitive text like quizzes, forms, or invoices and weighing regex vs LLM').

Broaden trigger-term coverage with natural user phrases such as 'parse forms', 'extract from invoices', 'regex vs LLM', 'hybrid extraction pipeline'.

Name a couple of concrete downstream actions (extract fields, validate low-confidence items) to reach the score-3 specificity anchor.

Dimension	Reasoning	Score
Specificity	Names a concrete action (choosing between Regex and LLM) and the domain (structured text parsing), but it is a single composite decision rather than multiple distinct concrete actions as the score-3 anchor requires.	2 / 3
Completeness	Clearly answers 'what' (a decision framework for regex-vs-LLM selection) but provides no explicit 'when/Use when' trigger clause, so per guideline completeness is capped at 2.	2 / 3
Trigger Term Quality	Contains relevant natural keywords ('正则表达式/Regex', '大型语言模型/LLM', '结构化文本', '边界情况') but lacks common user-speech variations and is not framed as explicit 'Use when' triggers.	2 / 3
Distinctiveness Conflict Risk	The regex-vs-LLM cost-optimization niche for structured text is specific and distinctive, with triggers unlikely to overlap with other skills.	3 / 3
	Total	9 / 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: xu-xiang/everything-claude-code-zh
Path: skills/regex-vs-llm-structured-text/SKILL.md
Commit: dfbf946

Reviewed: 8 minutes 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.