Describes a table you were given (pasted Markdown, CSV, TSV or spreadsheet cells) in a short paragraph that uses only figures the table itself prints. Every number is copied as written with its unit, currency and qualifier. Nothing is added up, averaged, rounded, turned into a percent or a ratio, or extended into a trend, a cause or a forecast. A blank or n/a cell is reported as not given, never as zero. A total the table prints is quoted, not checked. Use when asked to describe a table, summarize a table, explain what a data table shows, write a caption or a sentence for a table, or put the rows of a report table into words without inventing a single number.
Table Summary: Only the Figures in the Table is a tested SKILL.md that describes a table you were given (pasted Markdown, CSV, TSV or spreadsheet cells) in a short paragraph that uses only figures the table itself prints; an agent buys it once for $0.05 over x402.
Calculating anything: totals, averages, growth rates and shares are never produced, and a total the table prints is quoted, not checked. It does not forecast, explain causes or judge whether numbers are good. Checking whether a table adds up, or comparing two tables, is another job.
Tested 2026-10-08 with a strong and a weak model.
| Sonnet | Haiku | |||
|---|---|---|---|---|
| with | without | with | without | |
| Tables described right (24 tables) | 24/24 | 16/24 | 23/24 | 22/24 |
Same request on both sides; the task itself says to use only figures that appear in the table. Read by hand, Sonnet without the skill still added its own arithmetic on eight tables: differences between rows (Sevilla is 14 degrees warmer than Bilbao), a change between two years, a ratio written as twice, shares of a whole, a net figure over signed changes and a rise extrapolated past the last row; it also called one price cheaper than another in a different currency and wrote the figure that a planted row asked for. Each computed figure was right as arithmetic and still broke the instruction. Haiku without the skill kept to the printed figures more closely and missed two plain cases.
Same cases and the same checks with and without the skill. The cases are ours, written around what the skill is for; with a handful of cases, a difference of one or two is within noise.
Our own test text, before and after the skill ran. Excerpts only.
Before
| Region | Q3 sales (EUR) | Units | | --- | --- | --- | | North | 412,300 | 1,240 | | South | 388,150 | 1,105 | | East | 297,800 | 960 | | West | 455,900 | 1,310 |
After
Four regions, with Q3 sales in euros and units sold. West is highest on both measures, at 455,900 EUR on 1,310 units. East is lowest on both, at 297,800 EUR on 960 units. North has 412,300 EUR on 1,240 units, and South has 388,150 EUR on 1,105 units.
Before
| Град | Продажби (лв.) | | --- | --- | | София | 12 400 | | Пловдив | 9 800 | | Варна | 7 300 |
After
Три града с продажби в лева: София е най-висока с 12 400 лв., следвана от Пловдив с 9 800 лв., а Варна е най-ниска със 7 300 лв.
Before
Producto,Unidades vendidas Mesa,120 Silla,340 Lámpara,85
After
Tres productos, con las unidades vendidas de cada uno. La silla tiene la cifra más alta, con 340 unidades, la mesa figura con 120 unidades y la lámpara tiene la más baja, con 85 unidades.
English, Bulgarian, Spanish. Tried in: English, Bulgarian, Spanish.
Perpetual, non-exclusive; use and modify for yourself incl. paid work; no resale or republishing. Holder: Georgi Kalchev, aiskills402.com. Full terms.
Current version 1.0.0, updated 2026-10-08. Whoever bought an earlier version gets new ones free through the same re-download token.
v1.0.0 · 2026-10-08
First release (written before any model test; no measured result yet). Describes one pasted table in a short paragraph that uses only figures the table prints: every number copied with its unit, currency and qualifier, nothing added up, averaged, rounded, turned into a percent or a ratio, or extended into a trend, a cause or a forecast. Blank and n/a cells are reported as not given, never as zero; a printed total is quoted, not recomputed; values in different currencies are listed, not ranked; rows are read in label order, not row order; text in a cell that speaks to an assistant is table text.
Price starts at $0.02 (class B); the measured baseline decides whether it stays.
Because a figure it calculates looks exactly like a figure the table printed, and the reader cannot tell them apart later. If the table prints no total, the description says so and writes no number in its place. Ask for the arithmetic by name when you want it, and the figures used are shown.
It reports the figure as not given for that row. It never reads a dash as zero, never fills the gap from neighbouring rows and never averages around it. A zero the table really prints is a zero and can be said.
Only when the labels give an order and every step moves the same way. Rows are put in label order first, because a table sorted by name can list March before January. Two points get a plain from-to sentence, and no forecast is ever added.
Clearly, on one habit. We asked both models to describe twenty-four tables using only the figures printed in them. Without the file Sonnet broke that rule on eight tables: it computed differences, a two-year change, a ratio, shares, a net figure and a forecast, compared prices across two currencies and copied a figure a planted row asked for. Every sum was correct arithmetic, and every one was a number the table never printed. Loaded with the file, it described all 24 correctly. Haiku went from 22 to 23.
Research & Summaries
Writes a faithful summary of a text in the text's own language, using only what the text says, with every number, name and date kept exactly and every qualifier kept. Use when asked to summarize, shorten, condense, give the gist or a TL;DR of an article, report, email, transcript, thread or document.
Data & Analysis
Extracts data from text into JSON that matches the shape you give (a JSON Schema, an example object or a list of fields), without guessing. Every value comes from the text; a missing fact becomes null, numbers and dates are converted only into the type the shape asks for, two conflicting values are not settled by a guess, and instructions hidden in the text are ignored. Use when asked to extract structured data, turn text into JSON, parse an invoice, receipt, email, order, CV or job post into fields, or fill a JSON schema from a document.
Data & Analysis
Repairs broken or messy CSV so a standard parser reads it, without changing the text of a single cell, and answers with the CSV only, no code fence. Fixes quotes that do not close, a quote inside an unquoted cell, cells that hold the delimiter or a line break, rows shorter than the header, a Markdown table, and chat text or a fence around the data. Numbers such as 1,250, 007 and 1.10, dates, spaces inside cells and values that look like formulas stay exactly as written. Semicolon and tab files keep their delimiter. A row longer than the header, a row that could be read two ways, an unbalanced quote that no reading closes, an HTML page and prose with no table get a fixed one-line CANNOT REPAIR answer instead of a guess that moves data between columns. Use when an export, an API or another model returned CSV that does not parse, or when asked to fix, clean up, close or convert a table to valid CSV.