guidance

Control LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows with Guidance - Microsoft Research's constrained generation framework

Quality

70%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./backend/cli/skills/llm-tools/guidance/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.

The skill is highly actionable with extensive executable examples, but it is verbose and over-inlined: repeated boilerplate, duplicated patterns, and concept explanations inflate the body, and material that belongs in the reference files is kept inline rather than pointed to. Tightening the body and pushing detail into the existing references would improve both conciseness and progressive disclosure.

Suggestions

Factor the repeated model-setup boilerplate into a single snippet and reference it, and remove the duplicated ReAct agent (it appears in both Core Concepts #5 and Pattern 4).

Move the full pattern catalog and best-practice examples into references/examples.md, keeping only a concise quick-start and links in SKILL.md to improve progressive disclosure.

Trim explanatory prose ('How it works', 'Benefits', 'Use cases') that restates concepts Claude already knows about tokenization and constrained generation.

Dimension	Reasoning	Score
Conciseness	The body is mostly code-driven and useful, but is noticeably padded: the model-setup boilerplate (models.Anthropic('claude-sonnet-4-5-20250929')) is repeated ~10 times, the ReAct agent appears twice, and 'How it works'/'Benefits'/'Use cases' prose explains concepts Claude already knows.	3 / 5
Actionability	Abundant copy-paste-ready, executable code covering the common cases (regex/select constraints, chat context managers, guidance functions, classification, extraction, ReAct agent); specific examples dominate over abstract description.	5 / 5
Workflow Clarity	Content is organized as a feature reference and pattern catalog rather than a sequenced workflow; individual patterns have internal sequence but there are no explicit validation checkpoints or error-recovery feedback loops, and no overarching process to follow.	3 / 5
Progressive Disclosure	Real one-level-deep references exist (constraints.md, backends.md, examples.md) and are signaled in 'See Also', but the SKILL.md body inlines ~400 lines of core concepts, five full patterns, best practices, a comparison table, and performance notes that largely belong in those reference files.	3 / 5
	Total	14 / 20 Passed

Description

75%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, action-oriented, and clearly distinguishes Guidance from related skills, but it omits any explicit 'when to use' trigger guidance. Adding a 'Use when...' clause would raise completeness and trigger discoverability.

Suggestions

Add an explicit trigger clause such as 'Use when you need to guarantee valid JSON/XML/code output, constrain LLM generation with regex or grammars, or build multi-step generation workflows.'

Include more natural synonyms a user might say (e.g. 'structured output', 'output validation', 'format enforcement') to broaden trigger coverage.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'Control LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows' — giving comprehensive coverage of the framework's capabilities.	5 / 5
Completeness	The 'what' is clearly and concretely stated, but there is no 'Use when...' clause or equivalent explicit trigger guidance; per the rubric this caps completeness at 3.	3 / 5
Trigger Term Quality	Includes strong natural technical terms users would say ('regex', 'grammars', 'valid JSON/XML', 'structured formats', 'multi-step workflows'), but is missing some common synonyms and standalone 'validation' phrasing.	4 / 5
Distinctiveness Conflict Risk	Names the specific framework ('Guidance - Microsoft Research's constrained generation framework') with a clear niche and distinct triggers, giving minimal conflict risk with other skills.	5 / 5
	Total	17 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (574 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/llm-tools/guidance/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.