CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

write-better

Improve, rewrite, edit, or review prose so it is clear, direct, useful, and natural without inventing facts or erasing the author's voice. Use when drafting or revising emails, documentation, reports, essays, messages, UI copy, explanations, or other prose; when the user asks to write better, humanize text, remove AI-sounding language, tighten wording, improve clarity or tone, or make an action easier to understand. Do not use for translation or factual research unless writing quality is also part of the request.

71

Quality

89%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

85%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.

An unusually disciplined style guide: every rule is concrete, paired with Weak/Better examples, and anchored by a multi-pass editing workflow with a final validation checklist. The main gaps are cross-section redundancy and a monolithic ~600-line body whose phrase reference lists could be split into a references file for better progressive disclosure.

Suggestions

Move the long phrase and word blacklists (warning-word list in section 4, participial-phrase list in 17, promotional list in 19, politeness list in 22) into a single references/phrase-blacklists.md file, keeping one or two illustrative examples inline, so the SKILL.md body carries only the core rules.

Merge overlapping sections to reduce redundancy: combine sections 11 and 12 (both about repetition) and sections 9 and 10 (both about manufactured emphasis), and fold the emphasis rules of section 21 into section 10.

Compress the Editing process passes into a single checklist table (pass, what to check, what to replace) to cut roughly half of the process section's tokens while preserving the sequence.

DimensionReasoningScore

Conciseness

The body is dense and directive with almost no metadiscourse or explanation of concepts Claude already knows; nearly every line is a rule, a phrase blacklist, or a Weak/Better pair that earns its place. It falls short of 5 because several sections overlap (repetition is covered in sections 2, 11, and 12; manufactured emphasis in 9, 10, and 21) and could be tightened or merged.

4 / 5

Actionability

For an instruction-only skill the guidance is fully actionable: concrete word substitutions ("'use' instead of 'utilize'"), specific phrases to avoid ("I hope this message finds you well"), Weak/Better before-after pairs for each rule, and an 8-question final reader test. Specific examples cover the common prose problems, matching the top anchor without needing code.

5 / 5

Workflow Clarity

The 'Editing process' section defines six sequenced passes (meaning, structure, language, rhetoric, rhythm/punctuation, reader test), each with an explicit check list, ending in a numbered validation checklist and a clear deliverable instruction ("Return the final version. Do not expose drafts..."). This matches the clear-sequence-with-explicit-checklists anchor; the task involves no destructive or batch operations, so no validation cap applies.

5 / 5

Progressive Disclosure

Sections are well-organized with numbered headings and easy in-file navigation, but at ~600 lines the body inlines material that could live in reference files (the long phrase and word blacklists in sections 4, 17, 19, 22) with no bundle files used at all. This fits the 'some structure, content that should be separate is inline' anchor better than the 4 anchor, which expects most content appropriately split.

3 / 5

Total

17

/

20

Passed

Description

91%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.

A strong description: concrete actions and quality criteria, explicit 'Use when' triggers with natural phrasing, and a clear negative boundary. The only weaknesses are synonym-heavy verbs and broad genre scope creating minor overlap risk.

DimensionReasoningScore

Specificity

The description lists concrete actions ("Improve, rewrite, edit, or review prose") plus specific quality outcomes ("clear, direct, useful, and natural without inventing facts or erasing the author's voice"). It falls just short of the 5 anchor because the four verbs are near-synonyms of one core action rather than distinct capabilities like 'extract text, fill forms, merge documents'.

4 / 5

Completeness

It explicitly answers both what ("Improve, rewrite, edit, or review prose so it is clear, direct, useful, and natural...") and when ("Use when drafting or revising... when the user asks to write better, humanize text..."), and adds an explicit exclusion ("Do not use for translation or factual research unless writing quality is also part of the request"), which exceeds the 5 anchor.

5 / 5

Trigger Term Quality

It covers natural trigger phrases users actually say: "write better", "humanize text", "remove AI-sounding language", "tighten wording", "improve clarity or tone", plus concrete artifact types ("emails, documentation, reports, essays, messages, UI copy"). This matches the comprehensive-synonyms anchor; no meaningful natural terms are missing.

5 / 5

Distinctiveness Conflict Risk

The humanize/AI-sounding-language triggers and the translation/research exclusion give it a distinct niche with low conflict risk. However, it spans all prose genres, leaving minor overlap risk with more specialized writing skills (commit messages, technical docs), which keeps it just below the minimal-conflict 5 anchor.

4 / 5

Total

18

/

20

Passed

Validation

93%

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

CriteriaDescriptionResult

skill_md_line_count

SKILL.md is long (607 lines); consider splitting into references/ and linking

Warning

Total

15

/

16

Passed

Repository

branben/school-core

Path

.agents/skills/write-better/SKILL.md

Reviewed

Table of Contents

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.