[{"data":1,"prerenderedAt":570},["ShallowReactive",2],{"blog-article-\u002Fblog\u002Fskill-creator-v2-agent-skills-vs-code":3,"blog-surround-\u002Fblog\u002Fskill-creator-v2-agent-skills-vs-code":561},{"id":4,"title":5,"body":6,"canonical":545,"date":546,"description":12,"extension":547,"featured":548,"image":549,"meta":550,"navigation":551,"ogImage":549,"path":552,"provider":553,"published":551,"seo":554,"stem":555,"tags":556,"url":553,"__hash__":560},"blog\u002Fblog\u002Fskill-creator-v2-agent-skills-vs-code.md","I Used Skill Creator v2 to Improve One of My Agent Skills in VS Code",{"type":7,"value":8,"toc":526},"minimark",[9,13,16,19,41,44,47,52,55,58,75,78,81,85,93,96,99,102,105,109,112,115,119,122,125,128,131,151,154,165,168,171,174,178,181,184,187,190,194,197,200,223,226,229,233,236,239,242,245,248,251,254,257,261,264,267,270,273,281,284,287,301,304,308,311,314,317,325,328,331,334,338,341,344,347,350,353,361,365,368,371,388,391,394,397,401,404,407,410,413,416,433,436,439,442,446,449,452,455,472,475,478,481,485,488,499,502,505,508,511,514,518,523],[10,11,12],"p",{},"I just published a video showing how I used Skill Creator v2 to improve an existing AI skill inside VS Code, and honestly, I was seriously surprised at how much this thing does.",[10,14,15],{},"What impressed me most is that it does much more than just rewrite instructions.",[10,17,18],{},"It can:",[20,21,22,26,29,32,35,38],"ul",{},[23,24,25],"li",{},"review an existing skill",[23,27,28],{},"suggest targeted improvements",[23,30,31],{},"run evals against a baseline",[23,33,34],{},"compare outputs side by side",[23,36,37],{},"generate benchmark summaries",[23,39,40],{},"help optimize descriptions for better triggering",[10,42,43],{},"In the video, I ran it against a skill I had already created to see whether the updated version actually performed better, or if there was anything I was missing.",[10,45,46],{},"And it was a ton of fun.",[48,49,51],"h2",{"id":50},"the-skill-i-tested","The Skill I Tested",[10,53,54],{},"The skill I used for the demo was a skill I had already created called README Wizard.",[10,56,57],{},"It basically generates a polished, professional README for any project and is meant to kick in whenever someone mentions:",[20,59,60,63,66,69,72],{},[23,61,62],{},"improving a README",[23,64,65],{},"project documentation",[23,67,68],{},"badges",[23,70,71],{},"first impressions for a repo",[23,73,74],{},"making a GitHub repo look more professional",[10,76,77],{},"It also checks project metadata, reads best practices, uses badges, Mermaid diagrams, and works from a README template. (need to create a video for this too.. on it)",[10,79,80],{},"So rather than creating a skill from scratch, I wanted to see if Skill Creator v2 could improve something real that I had already built.",[48,82,84],{"id":83},"finding-skill-creator-v2","Finding Skill Creator v2",[10,86,87,88,92],{},"The first thing I did was go to ",[89,90,91],"code",{},"skills.sh"," and search for Anthropic.",[10,94,95],{},"From there, I found Skill Creator, and it now shows a summary of the skill which is nice.",[10,97,98],{},"The skill covers test case creation, evaluation and also runs parallel test cases with and without the skill to measure impact, capturing timing and token usage for comparison.",[10,100,101],{},"And on top of that, it generates an interactive browser-based reviewer showing outputs, qualitative feedback, and benchmark metrics.",[10,103,104],{},"It also includes description optimization, which is really important for improving skill triggering accuracy by testing realistic trigger and non-trigger queries.",[48,106,108],{"id":107},"installing-the-skill","Installing the Skill",[10,110,111],{},"Installing it was pretty straightforward.",[10,113,114],{},"I copied the install command from the page, pasted it into the terminal, and then selected where I wanted the skill installed.",[48,116,118],{"id":117},"looking-inside-the-skill","Looking Inside the Skill",[10,120,121],{},"Before running it, I wanted to see what was actually inside the Skill Creator skill.",[10,123,124],{},"Skills are written for agents, not really for you to sit there and read through line by line, but I always like to have a look.",[10,126,127],{},"And this one is pretty complex.",[10,129,130],{},"It includes:",[20,132,133,136,139,142,145,148],{},[23,134,135],{},"the SKILL.md file",[23,137,138],{},"its own agents",[23,140,141],{},"references and schemas",[23,143,144],{},"Python scripts",[23,146,147],{},"an eval viewer",[23,149,150],{},"review tooling",[10,152,153],{},"What I found really cool is that it comes with its own agents:",[20,155,156,159,162],{},[23,157,158],{},"an analyzer agent",[23,160,161],{},"a comparator",[23,163,164],{},"a grader",[10,166,167],{},"The analyzer looks at comparison results and tries to understand why the winner won and generate better suggestions.",[10,169,170],{},"The comparator compares two outputs without knowing which skill produced them.",[10,172,173],{},"The grader evaluates expectations against the execution transcript and outputs.",[48,175,177],{"id":176},"running-it-against-a-real-skill","Running It Against a Real Skill",[10,179,180],{},"I then used Skill Creator against my README Wizard skill.",[10,182,183],{},"I’ve done this a couple of times now, and I found that in VS Code I sometimes need to be a little more explicit if I want the full benefit of the sub agents.",[10,185,186],{},"Claude seems to pick that up more naturally because the skill was built for it, but in VS Code I wanted to make sure it really used everything available.",[10,188,189],{},"So I’d definitely encourage being explicit there.",[48,191,193],{"id":192},"what-it-found","What It Found",[10,195,196],{},"Very quickly, it started identifying issues with my skill.",[10,198,199],{},"Things like:",[20,201,202,205,208,211,214,217,220],{},[23,203,204],{},"the workflow being under-specified",[23,206,207],{},"missing guidance for handling existing or missing READMEs",[23,209,210],{},"README best practices being too thin",[23,212,213],{},"sections that should only appear if relevant links exist",[23,215,216],{},"eval coverage being too small",[23,218,219],{},"missing edge cases",[23,221,222],{},"limited project detection",[10,224,225],{},"For example, it pointed out that a personal learning repo probably doesn’t need the same sections as every other project.",[10,227,228],{},"It also spotted that I only had two evals and suggested adding more realistic test cases, including edge cases like minimal projects and badge-focused README requests.",[48,230,232],{"id":231},"applying-improvements","Applying Improvements",[10,234,235],{},"Once it had reviewed everything, it started applying targeted improvements across the skill files.",[10,237,238],{},"This part was honestly kind of exciting to watch because it moved fast.",[10,240,241],{},"It updated the skill instructions and made the guidance more explicit.",[10,243,244],{},"It improved the best practices.",[10,246,247],{},"It tightened up the logic around when certain sections should or shouldn’t be included.",[10,249,250],{},"And it expanded the eval coverage.",[10,252,253],{},"I could go through the changed files while it was working and see that it wasn’t just randomly changing things. It was making focused improvements that actually made sense.",[10,255,256],{},"That part gave me a lot of confidence.",[48,258,260],{"id":259},"adding-more-evals","Adding More Evals",[10,262,263],{},"One thing I really liked was how it expanded the eval set.",[10,265,266],{},"I had two evals.",[10,268,269],{},"It added more.",[10,271,272],{},"For example, it created cases around:",[20,274,275,278],{},[23,276,277],{},"minimal project README generation",[23,279,280],{},"badge-focused requests",[10,282,283],{},"And these evals work like tests.",[10,285,286],{},"They include assertions such as whether the output has:",[20,288,289,292,295,298],{},[23,290,291],{},"a project description",[23,293,294],{},"a quick start or usage section",[23,296,297],{},"appropriate badges",[23,299,300],{},"the right structure for the kind of project being documented",[10,302,303],{},"This was super useful because it meant I wasn’t just guessing whether the skill was better. I could actually measure it.",[48,305,307],{"id":306},"running-sub-agents-in-parallel","Running Sub Agents in Parallel",[10,309,310],{},"Then came one of the coolest parts.",[10,312,313],{},"It launched sub-agent runs in parallel.",[10,315,316],{},"It ran:",[20,318,319,322],{},[23,320,321],{},"the improved skill",[23,323,324],{},"the old skill baseline",[10,326,327],{},"side by side across multiple test cases.",[10,329,330],{},"That meant it could directly compare the version with the new changes against the original version.",[10,332,333],{},"This is where the workflow really stood out to me. It wasn’t just making edits and calling it a day. It was actually testing whether the changes improved results.",[48,335,337],{"id":336},"the-results","The Results",[10,339,340],{},"After the runs completed, it graded all the outputs against their assertions and generated benchmark results.",[10,342,343],{},"The improved skill outperformed the baseline on two out of four evals and tied on the other two.",[10,345,346],{},"The overall result improved from 81 to 97.5.",[10,348,349],{},"That’s a 20.4% improvement.",[10,351,352],{},"Some of the biggest wins came from improving the skill’s ability to:",[20,354,355,358],{},[23,356,357],{},"generate good content even when metadata is sparse",[23,359,360],{},"adapt README length and sections to different project types instead of always forcing the full template",[48,362,364],{"id":363},"the-workspace-it-creates","The Workspace It Creates",[10,366,367],{},"Another thing I wanted to show in the video was the workspace it creates while doing all this.",[10,369,370],{},"It creates a workspace folder where it stores things like:",[20,372,373,376,379,382,385],{},[23,374,375],{},"skill snapshots",[23,377,378],{},"old skill outputs",[23,380,381],{},"grading results",[23,383,384],{},"benchmark data",[23,386,387],{},"iteration files",[10,389,390],{},"You don’t necessarily need to go through all of that manually, but it’s very cool that you can.",[10,392,393],{},"If you want to inspect exactly what happened at each stage, it’s all there.",[10,395,396],{},"That level of visibility is really nice.",[48,398,400],{"id":399},"the-html-eval-viewer","The HTML Eval Viewer",[10,402,403],{},"Then I asked whether there was a way to see the benchmarks in HTML.",[10,405,406],{},"And yes, there is.",[10,408,409],{},"Skill Creator has an eval viewer for that.",[10,411,412],{},"This was another really nice surprise.",[10,414,415],{},"It launched an HTML review page showing:",[20,417,418,421,424,427,430],{},[23,419,420],{},"old skill vs improved skill",[23,422,423],{},"formal grades",[23,425,426],{},"pass\u002Ffail results",[23,428,429],{},"benchmark comparisons",[23,431,432],{},"review flows for feedback submission",[10,434,435],{},"It’s made for a human to read.",[10,437,438],{},"You can actually review what happened and decide whether you agree with the results.",[10,440,441],{},"I really liked that.",[48,443,445],{"id":444},"description-optimization","Description Optimization",[10,447,448],{},"And then, because apparently this skill wasn’t done showing off yet, I ran the description optimization flow as well.",[10,450,451],{},"This generates trigger and non-trigger queries to see whether your skill description is actually good enough to fire when it should, and stay out of the way when it shouldn’t.",[10,453,454],{},"That workflow lets you:",[20,456,457,460,463,466,469],{},[23,458,459],{},"review trigger queries",[23,461,462],{},"review non-trigger queries",[23,464,465],{},"edit them",[23,467,468],{},"export the eval set",[23,470,471],{},"run the optimization loop",[10,473,474],{},"That is super valuable.",[10,476,477],{},"A lot of the time, the problem with a skill is not the logic inside it. It’s that the description is not specific enough, or not clear enough, for the agent to trigger it properly.",[10,479,480],{},"So I really liked that this was built in too.",[48,482,484],{"id":483},"final-thoughts","Final Thoughts",[10,486,487],{},"If you’re already building custom skills for:",[20,489,490,493,496],{},[23,491,492],{},"GitHub Copilot",[23,494,495],{},"Claude Code",[23,497,498],{},"or other coding agents",[10,500,501],{},"this is absolutely worth checking out.",[10,503,504],{},"And yes, the video is long.",[10,506,507],{},"But the skill does a lot, and I found it hard to cut it down because I kept finding more things it could do.",[10,509,510],{},"So I pretty much left it as is.",[10,512,513],{},"Have fun.",[48,515,517],{"id":516},"watch-the-video","Watch the Video",[519,520],"lite-youtube",{"videoid":521,"playlabel":522},"WplS5lycPHM","Build Better AI Agent Skills With Skill Creator v2 from Anthropic",[10,524,525],{},"If you’re creating your own skills already, or even just experimenting with prompts and instructions, I’d be really curious to know how you’re approaching it.",{"title":527,"searchDepth":528,"depth":528,"links":529},"",2,[530,531,532,533,534,535,536,537,538,539,540,541,542,543,544],{"id":50,"depth":528,"text":51},{"id":83,"depth":528,"text":84},{"id":107,"depth":528,"text":108},{"id":117,"depth":528,"text":118},{"id":176,"depth":528,"text":177},{"id":192,"depth":528,"text":193},{"id":231,"depth":528,"text":232},{"id":259,"depth":528,"text":260},{"id":306,"depth":528,"text":307},{"id":336,"depth":528,"text":337},{"id":363,"depth":528,"text":364},{"id":399,"depth":528,"text":400},{"id":444,"depth":528,"text":445},{"id":483,"depth":528,"text":484},{"id":516,"depth":528,"text":517},"https:\u002F\u002Fdev.to\u002Fdebs_obrien\u002Fi-used-skill-creator-v2-to-improve-one-of-my-agent-skills-in-vs-code-fhd","2026-03-21","md",false,"\u002Fimages\u002Fblog\u002Fcovers\u002Fskill-creator-v2-agent-skills-vs-code.png",{},true,"\u002Fblog\u002Fskill-creator-v2-agent-skills-vs-code",null,{"title":5,"description":12},"blog\u002Fskill-creator-v2-agent-skills-vs-code",[557,558,559],"ai","githubcopilot","webdev","XZFR-VBcPD-vpZygK_wC4BoYhvAF75KBQnEloetuNkc",[562,566],{"title":563,"path":564,"stem":565,"children":-1},"I Built an Agent Skill That Turns Rough READMEs Into Polished Docs","\u002Fblog\u002Fbuilt-agent-skill-readme-wizard","blog\u002Fbuilt-agent-skill-readme-wizard",{"title":567,"path":568,"stem":569,"children":-1},"Build Websites, Games, and Teaching Resources With Google Gemini for Free (No Coding Required)","\u002Fblog\u002Fbuild-websites-games-and-teaching-resources-with-google-gemini","blog\u002Fbuild-websites-games-and-teaching-resources-with-google-gemini",1793092821306]