agentsclimarketplace

Readability

Skill dzivkovi/agent-skills-on-databricks/skills/readability

Bring your Claude Code skill unchanged and run it as a governed Databricks job. Publish once to a Unity Catalog volume, reuse from any job, chain skills into a pipeline (markdown in, branded PowerPoint out). No external API key. Runs on Free Edition.

Install
npx -y skills add dzivkovi/agent-skills-on-databricks --skill readability

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 22 days oldThe repository was created 22 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Grade how readable a text document is by pairing EXACT, code-computed readability metrics (Flesch Reading Ease, Flesch-Kincaid grade level, syllable and hard-word counts) with an LLM's plain-language coaching (who can read this, and the two edits that would lower the grade most). Use when you need a grounded readability assessment where the scores must be correct, not estimated. Triggers: 'how readable is this', 'reading level', 'grade level', 'make this easier to read', 'plain language check'.

SKILL.md

3.0 KB, as published. Nobody here has run it

Readability

The second skill in this repo, and proof that one runner serves many skills: same two-half shape as document-insights (deterministic facts + LLM judgment), but a DIFFERENT contract. Here the exact half computes readability scores; the LLM half coaches how to improve them.

How to run this skill

  1. Deterministic step - run scripts/analyze.py <input-file>. It returns a JSON object of exact metrics: word_count, sentence_count, syllable_count, avg_syllables_per_word, flesch_reading_ease, flesch_kincaid_grade, hard_word_count, longest_sentence_words. These scores are ground truth. Never let the LLM recompute or "estimate" a grade.
  2. Non-deterministic step - ask the model for a plain-language read of the scores: name the likely audience (e.g. "general adult reader", "needs a college reading level"), and give the two concrete edits that would most lower the Flesch-Kincaid grade (shorten the longest sentences, swap the multi-syllable words). Ground every claim in the metrics and the document; never invent a score.

Output contract

The report is built as two sections, whose headings and metrics table are owned by the caller: a Metrics (computed by code - exact) section (the exact analyze.py numbers as a metric-value table) followed by a Reading (interpreted by the LLM) section. Your job is only the body of that Reading section - no headings, no table. In it:

  • open with the likely audience and what the Flesch Reading Ease band means (90-100 very easy, 60-70 plain English, 30-50 difficult, 0-30 very difficult), tied to the exact score;
  • then give the top two edits that would most lower the Flesch-Kincaid grade (shorten the longest sentences, swap the multi-syllable words), each tied to a specific metric.

Use hyphens, not em-dashes. Keep it concise.

Why this split matters

An LLM asked "what grade level is this?" will guess. analyze.py computes the Flesch scores exactly from syllables and sentence length. The LLM cannot count syllables reliably, but it is good at turning a grade into actionable coaching - so each half does what it is good at, and the output labels which is which.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.