--- title: Agentic Tip #5: Write skills after the fact | daily.dev description: A practical tip for working with AI agents: instead of writing reusable skills upfront before a workflow is proven, run the workflow manually with the agent... canonical: https://daily.dev/posts/agentic-tip-5-write-skills-after-the-fact-ajlu6oxeg twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Agentic Tip #5: Write skills after the fact | daily.dev og:description: A practical tip for working with AI agents: instead of writing reusable skills upfront before a workflow is proven, run the workflow manually with the agent... og:url: https://daily.dev/posts/agentic-tip-5-write-skills-after-the-fact-ajlu6oxeg og:image: https://api.daily.dev/og/posts/aJLU6oxEg.png og:image:alt: Agentic Tip #5: Write skills after the fact og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Agentic Tip #5: Write skills after the fact **[Agentic Engineering](https://daily.dev/sources/agentic)** · [@idoshamun](https://daily.dev/idoshamun) · 1 min read · 23 upvotes · 1 comments ## Summary A practical tip for working with AI agents: instead of writing reusable skills upfront before a workflow is proven, run the workflow manually with the agent first, refine it step by step, then ask the agent to convert the conversation into a reusable skill. This approach increases the likelihood the skill works correctly and retains all necessary context. ## Content You're about to run a new workflow with your agent, one you already know you'll run a million times. It's so tempting to go ahead and write a skill right away. But most likely that skill will suck and won't work on the first try. Instead, my recommendation after many iterations: run the flow yourself with the agent first. One prompt at a time. Perfect the process, make sure it delivers what you want. Only then, at the end of the discussion, tell the agent: "turn this conversation into a skill I can reuse in the future." It's not bulletproof, but it makes it far more likely the skill actually works and carries all the context it needs. --- In case you missed, [tip #4 is here.](https://daily.dev/posts/4aVT5DxsD) ## Community discussion Top comments from developers on daily.dev. **@pdfopsdev** · 0 upvotes > This matches my experience hard. The reason retro-authored skills beat upfront ones is that the manual run captures the failure modes and recovery branches you'd never think to specify in the abstract — the transcript is a worked example with all the dead-ends still visible, so the skill encodes 'when X fails, do Y' instead of just the happy path. One addition: prune the transcript before converting. Leave the recoveries in but cut the noise, or the skill inherits your false starts as if they were the intended procedure. ## Similar posts on daily.dev - [Writing and Reviewing Agent Skills - Common Pitfalls](https://daily.dev/posts/writing-and-reviewing-agent-skills---common-pitfalls-liuwqgi1b) · Sam McLeod · 0 upvotes · 0 comments - [Beyond Prompting: Using Agent Skills in Data Science](https://daily.dev/posts/beyond-prompting-using-agent-skills-in-data-science-lcepbpr0c) · Towards Data Science · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#prompt-engineering](https://daily.dev/tags/prompt-engineering), [#vibe-coding](https://daily.dev/tags/vibe-coding) [View this post on daily.dev](https://daily.dev/posts/agentic-tip-5-write-skills-after-the-fact-ajlu6oxeg) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/agentic-tip-5-write-skills-after-the-fact-ajlu6oxeg","headline":"Agentic Tip #5: Write skills after the fact","text":"A practical tip for working with AI agents: instead of writing reusable skills upfront before a workflow is proven, run the workflow manually with the agent first, refine it step by step, then ask the agent to convert the conversation into a reusable skill. This approach increases the likelihood the skill works correctly and retains all necessary context.","url":"https://daily.dev/posts/agentic-tip-5-write-skills-after-the-fact-ajlu6oxeg","datePublished":"2026-07-10T05:30:00.215Z","dateModified":"2026-09-13T20:54:54.366Z","author":{"@type":"Person","name":"Ido Shamun","url":"https://daily.dev/idoshamun","image":"https://media.daily.dev/image/upload/s---xy_OAwk--/f_auto,q_auto/v1703781380/avatars/avatar_28849d86070e4c099c877ab6837c61f0","description":"I'm to blame if something goes wrong here 🙈","worksFor":{"@type":"Organization","name":"daily.dev","logo":"https://daily-now-res.cloudinary.com/image/upload/s--MASZOhnv--/f_auto/v1614088267/landing/Daily.dev_logo"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"EndorseAction"},"userInteractionCount":192210}},"image":"https://media.daily.dev/image/upload/s--lh9TfEa5--/f_auto/v1783340716/posts/aJLU6oxEg?_a=BAMAMicg0","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":23},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"comment":[{"@type":"Comment","text":"This matches my experience hard. The reason retro-authored skills beat upfront ones is that the manual run captures the failure modes and recovery branches you’d never think to specify in the abstract — the transcript is a worked example with all the dead-ends still visible, so the skill encodes ‘when X fails, do Y’ instead of just the happy path. One addition: prune the transcript before converting. Leave the recoveries in but cut the noise, or the skill inherits your false starts as if they were the intended procedure.","datePublished":"2026-07-10T18:06:45.283Z","url":"https://daily.dev/posts/aJLU6oxEg#c-cvrXtg42E","author":{"@type":"Person","name":"PDFops","url":"https://daily.dev/pdfopsdev","image":"https://media.daily.dev/image/upload/s---8isRBKc--/f_auto/v1782922291/avatars/avatar_orjMeK8QKaaVZwGq7ScPz?_a=BAMAMicg0"}}],"isPartOf":{"@type":"WebPage","url":"https://daily.dev/squads/agentic","name":"Agentic Engineering"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Agentic Engineering","item":"https://daily.dev/squads/agentic"},{"@type":"ListItem","position":3,"name":"Agentic Tip #5: Write skills after the fact"}]} ```