Blog post: a worked-example Agent Skill for Great Tables - #380
Open
HrudithL wants to merge 11 commits into
Open
Conversation
Compares three great_tables skill designs from an internal eval harness (gt-skill) against hand-built ground truths. The thin "house" design -- one worked reference script plus a short rules file, no flowchart -- comes out cheapest per table while landing within a point of the top-scoring design and posting the tightest run-to-run consistency. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Reworks the earlier draft per feedback: drops the multi-skill comparison in favor of presenting the skill on its own terms, swaps in a fresh eval sweep (round 4) and a clearer example (gtcars top-10-by-country, repeat 3 vs. baseline), combines the token and cost charts into one side-by-side image, regenerates all charts with generic labels, and rewrites the prose to read less like a comparison report. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
✅ Deploy Preview for posit-open-source canceled.
|
Contributor
Publishing checklist
|
Contributor
- Replace prose with a shorter, less-technical walkthrough of the skill - Swap old comparator/consistency/cost/ground-truth plots for the refreshed accuracy and usage plots from eval-results/house - Reuse existing no-skill-table and skill-table screenshots as the good/bad table examples - Regenerate index.md via quarto render
…he Great Tables Skill
Contributor
Blog YAML Checks
|
Contributor
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Prerequisite check
Related Issue
N/A
Description of changes
Adds a new blog post about a worked-example Agent Skill for Great Tables, evaluated against a hand-built ground-truth corpus. Covers why table design decisions matter, what changes with the skill loaded vs. not (using a real before/after example), how the skill stays lightweight (one reference script plus two short reference files, no numbered flowchart), the corpus-wide numbers (comparator score, run-to-run consistency, token/cost usage vs. baseline), and a side-by-side check against a hand-authored ground truth for one prompt.
Also adds a minimal
content/people/profile for the author.Screenshots or previews (if applicable)
See the Netlify preview link this PR generates for the full rendered post. The post's hero image (
featured.pngin the post folder) shows the core before/after comparison: the same prompt and data, once with no skill loaded and once with the skill attached.Checklist
CONTRIBUTING.mdguidelines.