[{"data":1,"prerenderedAt":1029},["ShallowReactive",2],{"guide-en-skills\u002Fclaude-skills-vs-mcp":3,"guide-siblings-en-skills":350,"guide-alt-en-skills\u002Fclaude-skills-vs-mcp":1026},{"id":4,"title":5,"author":6,"body":7,"date":324,"description":325,"extension":326,"faq":327,"meta":340,"navigation":341,"order":342,"path":343,"readTime":344,"seo":345,"stem":346,"topic":347,"translationId":348,"updated":324,"__hash__":349},"guides\u002Fguides\u002Fskills\u002Fclaude-skills-vs-mcp.md","Claude skills vs MCP: what each one does and when to use which","Walma Engineering",{"type":8,"value":9,"toc":312},"minimark",[10,14,26,31,44,61,65,174,178,184,187,190,193,197,200,207,238,241,245,260,263,275,278,282,285,289,292,299],[11,12,13],"p",{},"Two features arrived a year apart and get confused constantly. MCP, the Model Context Protocol, launched in November 2024. Skills launched in October 2025. Both extend what an agent can do, and both live in folders and config files, so it is natural to ask which one to use.",[11,15,16,17,21,22,25],{},"The short answer: MCP is about ",[18,19,20],"strong",{},"access",", skills are about ",[18,23,24],{},"know-how",". The rest of this page makes that precise.",[27,28,30],"h2",{"id":29},"one-sentence-each","One sentence each",[11,32,33,34,37,38,43],{},"An ",[18,35,36],{},"MCP server"," gives the agent tools it did not have: search GitHub, query a database, post to Slack, drive a browser. It is a program running somewhere, exposing functions through a standard protocol. Read ",[39,40,42],"a",{"href":41},"\u002Fen\u002Fguides\u002Fmcp","our MCP guide"," for the full picture.",[11,45,46,47,50,51,55,56,60],{},"A ",[18,48,49],{},"skill"," gives the agent instructions it did not have: how to write release notes the way your team does, how to produce a valid Excel file, how to run your incident checklist. It is a folder with a ",[52,53,54],"code",{},"SKILL.md"," and optional scripts. Read ",[39,57,59],{"href":58},"\u002Fen\u002Fguides\u002Fskills","our skills guide"," for the details.",[27,62,64],{"id":63},"side-by-side","Side by side",[66,67,68,82],"table",{},[69,70,71],"thead",{},[72,73,74,77,79],"tr",{},[75,76],"th",{},[75,78,36],{},[75,80,81],{},"Skill",[83,84,85,97,108,119,130,141,152,163],"tbody",{},[72,86,87,91,94],{},[88,89,90],"td",{},"What it adds",[88,92,93],{},"Tools, resources, prompts from an external system",[88,95,96],{},"Instructions, examples, scripts for a task",[72,98,99,102,105],{},[88,100,101],{},"Where it runs",[88,103,104],{},"A separate process, local or remote",[88,106,107],{},"Inside the agent's context; scripts run in the sandbox or shell",[72,109,110,113,116],{},[88,111,112],{},"Needs credentials",[88,114,115],{},"Usually (OAuth, API keys)",[88,117,118],{},"Rarely; uses whatever tools are already available",[72,120,121,124,127],{},[88,122,123],{},"Loaded when",[88,125,126],{},"Every session (tool descriptions)",[88,128,129],{},"Only when a task matches its description",[72,131,132,135,138],{},[88,133,134],{},"Token cost",[88,136,137],{},"Per session, proportional to number of tools",[88,139,140],{},"Zero until used, then the file length",[72,142,143,146,149],{},[88,144,145],{},"Portable to other agents",[88,147,148],{},"Yes, any MCP client",[88,150,151],{},"Yes, via the open Agent Skills spec",[72,153,154,157,160],{},[88,155,156],{},"Main risk",[88,158,159],{},"Over-permission, prompt injection via results",[88,161,162],{},"Untrusted instructions and scripts",[72,164,165,168,171],{},[88,166,167],{},"Owned by",[88,169,170],{},"Usually the vendor of the system",[88,172,173],{},"Usually your team",[27,175,177],{"id":176},"the-question-that-decides-it","The question that decides it",[11,179,180,181],{},"Ask: ",[18,182,183],{},"does the agent lack access, or lack knowledge?",[11,185,186],{},"If the agent cannot do the task because it cannot reach something (the ticket system, the data warehouse, the deployment pipeline), that is an access problem. You need an MCP server, or a CLI the agent can call.",[11,188,189],{},"If the agent can reach everything but does the task badly (wrong format, wrong order, forgets a step, ignores a convention), that is a knowledge problem. You need a skill.",[11,191,192],{},"Most real tasks are both. \"Prepare the release\" needs access to GitHub and the deployment system (MCP) and the knowledge of what your release procedure is (skill).",[27,194,196],{"id":195},"how-they-work-together","How they work together",[11,198,199],{},"Here is a typical workflow with both.",[11,201,202,203,206],{},"A developer types \"cut release 2.5\". The agent matches the request to a ",[52,204,205],{},"release"," skill and loads it. The skill says:",[208,209,210,214,221,228,231],"ol",{},[211,212,213],"li",{},"Confirm the main branch is green using the CI tool.",[211,215,216,217,220],{},"List merged PRs since the last tag using the GitHub tool, excluding ",[52,218,219],{},"chore",".",[211,222,223,224,227],{},"Draft release notes in the house format (example in ",[52,225,226],{},"examples\u002F",").",[211,229,230],{},"Create the tag and the GitHub release using the GitHub tool.",[211,232,233,234,237],{},"Post the notes to ",[52,235,236],{},"#releases"," using the Slack tool, then stop. Do not deploy.",[11,239,240],{},"Steps 1, 2, 4 and 5 call MCP tools. Steps 3 and the overall order come from the skill. Neither alone gets you a release done your way.",[27,242,244],{"id":243},"when-a-skill-is-enough","When a skill is enough",[11,246,247,248,251,252,255,256,259],{},"Skills can carry scripts, and scripts can call CLIs. If ",[52,249,250],{},"gh",", ",[52,253,254],{},"aws"," or ",[52,257,258],{},"kubectl"," are already installed and authenticated on the machine, a skill can use them through the shell without any MCP server. For personal use that is often simpler.",[11,261,262],{},"It stops being enough when:",[264,265,266,269,272],"ul",{},[211,267,268],{},"the agent runs somewhere without those CLIs (Claude.ai, the API, a CI runner);",[211,270,271],{},"you need per-user authorization rather than whatever the laptop's credentials allow;",[211,273,274],{},"you want the access to be discoverable and governable across a team.",[11,276,277],{},"At that point the access belongs in an MCP server.",[27,279,281],{"id":280},"when-mcp-is-enough","When MCP is enough",[11,283,284],{},"If the task is genuinely \"do the obvious thing with this tool\" (\"find open bugs assigned to me\", \"what changed in this file\"), tool descriptions carry enough guidance and no skill is needed. Skills earn their place when there is a procedure, a format or a convention the model would not guess.",[27,286,288],{"id":287},"governance-they-meet-at-the-gateway","Governance: they meet at the gateway",[11,290,291],{},"For a team, both need an approved list. Skills are code and instructions from somewhere; MCP servers are code with credentials. A reviewed, versioned catalogue of each is the minimum.",[11,293,294,295,298],{},"But there is an asymmetry worth understanding. A skill describes a procedure; it cannot enforce who may run it. An MCP tool actually performs the action; it can be gated per user. So the enforcement point for \"only the release manager may tag a release\" is not the release skill but the policy on the GitHub server's ",[52,296,297],{},"create_release"," tool.",[11,300,301,302,306,307,311],{},"That is why, in a company deployment, both end up behind the same ",[39,303,305],{"href":304},"\u002Fen\u002Fguides\u002Fmcp\u002Fmcp-gateway","gateway",": skills and servers are distributed through one client, tool policy is enforced per user, and the log shows which skill drove which tool call. Walma AI Hub does exactly that, inside the customer's own EU region. ",[39,308,310],{"href":309},"\u002Fen\u002Fai-hub","Book a walkthrough"," if you are working out how to standardise both across a team.",{"title":313,"searchDepth":314,"depth":315,"links":316},"",2,3,[317,318,319,320,321,322,323],{"id":29,"depth":314,"text":30},{"id":63,"depth":314,"text":64},{"id":176,"depth":314,"text":177},{"id":195,"depth":314,"text":196},{"id":243,"depth":314,"text":244},{"id":280,"depth":314,"text":281},{"id":287,"depth":314,"text":288},"2026-09-11","Skills teach an agent how to do a task. MCP gives it access to tools and data. They are not competitors. Here is a clear comparison, the questions that decide which you need, and how they work together in a real workflow.","md",[328,331,334,337],{"q":329,"a":330},"Should I use skills or MCP?","Usually both. Use MCP when the agent needs to reach a system it cannot otherwise access, such as GitHub, a database or a SaaS tool. Use a skill when the agent needs to know how to do a task well, such as your team's review process or a document format. A skill often uses MCP tools to do its work.",{"q":332,"a":333},"Can a skill replace an MCP server?","Only if the task needs no external access, or the access is available through the shell or a CLI already on the machine. A skill can wrap a CLI in a script. For anything that needs authentication to a remote system, an MCP server is the cleaner and safer option.",{"q":335,"a":336},"Which is more secure, skills or MCP?","They have different risks. MCP servers hold credentials and can act on systems, so their risk is over-permission and injection through results. Skills run scripts and shape the model's behaviour, so their risk is supply chain: an untrusted skill can instruct the agent to do harmful things. Both need review and an allowlist in a team setting.",{"q":338,"a":339},"Do skills use more tokens than MCP?","A skill costs nothing until it is loaded, then costs the length of its SKILL.md. An MCP server costs its tool descriptions on every session. A server with 40 tools can be more expensive per session than several skills.",{},true,1,"\u002Fguides\u002Fskills\u002Fclaude-skills-vs-mcp","7 min read",{"title":5,"description":325},"guides\u002Fskills\u002Fclaude-skills-vs-mcp","skills","claude-skills-vs-mcp","ngAyN_wBknv7_7oNwBwgYBMPoZ2RQmtrUDikk15jkaQ",[351,823],{"id":352,"title":353,"author":6,"body":354,"date":324,"description":798,"extension":326,"faq":799,"meta":815,"navigation":341,"order":816,"path":817,"readTime":818,"seo":819,"stem":820,"topic":347,"translationId":821,"updated":324,"__hash__":822},"guides\u002Fguides\u002Fskills\u002Findex.md","Claude skills explained: what they are, how to install them, and how to write your own",{"type":8,"value":355,"toc":787},[356,362,369,373,376,382,386,389,412,415,419,422,431,434,581,588,591,597,610,614,617,627,636,646,649,653,656,659,663,666,672,678,684,690,699,705,709,712,715,719,732,738,744,752,756,759,765,771,777,783],[11,357,358,359,361],{},"A skill is a folder. Inside it is a Markdown file called ",[52,360,54],{}," that tells Claude how to do a specific job, and optionally some scripts and reference documents that help. When a task matches the skill's description, Claude reads the file and follows it. When it does not, the skill costs nothing.",[11,363,364,365,220],{},"That is the whole mechanism, and it turns out to be the most practical way anyone has found to give an AI agent expertise that survives from one session to the next. This guide covers how skills work, how to install them in Claude Code, Claude.ai and the API, how to write one that actually improves results, and what to consider before a team adopts them. For how skills relate to MCP, see ",[39,366,368],{"href":367},"\u002Fen\u002Fguides\u002Fskills\u002Fclaude-skills-vs-mcp","Claude skills vs MCP",[27,370,372],{"id":371},"why-skills-exist","Why skills exist",[11,374,375],{},"Prompts do not scale. A good prompt for \"review this pull request the way our team does it\" is two pages long. Nobody pastes two pages into every session, so the knowledge lives in someone's head, or in a wiki the agent never reads.",[11,377,378,379,381],{},"Skills package that knowledge once, next to the code, and load it only when needed. Anthropic launched them in October 2025 for Claude.ai, Claude Code and the API. In December 2025 the format was published as an open specification, Agent Skills, and other coding agents adopted it. In practice, a ",[52,380,54],{}," you write for Claude Code today also works in Codex, Cursor and Copilot with minor differences.",[27,383,385],{"id":384},"how-a-skill-works","How a skill works",[11,387,388],{},"Three things happen at different times, which is why skills scale.",[208,390,391,397,406],{},[211,392,393,396],{},[18,394,395],{},"At startup",", the agent reads only the frontmatter of every skill it knows about: the name and a one-line description. For fifty skills that is a few hundred tokens.",[211,398,399,402,403,405],{},[18,400,401],{},"When a task matches a description",", the agent reads the full ",[52,404,54],{}," into context. That is when the instructions start to matter.",[211,407,408,411],{},[18,409,410],{},"When the instructions say so",", the agent opens reference files or runs scripts in the skill folder. A skill for generating Excel files might carry a Python script that does the actual writing, so the model does not have to reinvent it.",[11,413,414],{},"This is called progressive disclosure. It is the reason you can have many skills installed without paying for all of them every turn.",[27,416,418],{"id":417},"anatomy-of-a-skill","Anatomy of a skill",[11,420,421],{},"The minimum is one file:",[423,424,429],"pre",{"className":425,"code":427,"language":428},[426],"language-text","release-notes\u002F\n└── SKILL.md\n","text",[52,430,427],{"__ignoreMap":313},[11,432,433],{},"With this content:",[423,435,439],{"className":436,"code":437,"language":438,"meta":313,"style":313},"language-markdown shiki shiki-themes github-dark","---\nname: release-notes\ndescription: Write release notes from merged pull requests in our house style. Use when asked to draft, write or prepare release notes or a changelog for a version.\n---\n\n# Release notes\n\n## When to use\nThe user asks for release notes, a changelog or \"what shipped\" for a version or date range.\n\n## Steps\n1. List merged PRs for the range using the GitHub tools. Ignore PRs labelled `chore` or `internal`.\n2. Group by: New, Improved, Fixed. One line each, present tense, no PR numbers in the line.\n3. Lead with the change that affects the most users.\n4. End with an \"Upgrade notes\" section only if any PR is labelled `breaking`.\n\n## Style\nShort sentences. No exclamation marks. British spelling. Product names as in `docs\u002Fstyle.md`.\n","markdown",[52,440,441,449,455,460,465,471,477,482,488,494,499,505,527,536,545,559,564,570],{"__ignoreMap":313},[442,443,445],"span",{"class":444,"line":342},"line",[442,446,448],{"class":447},"sIZOC","---\n",[442,450,451],{"class":444,"line":314},[442,452,454],{"class":453},"s95oV","name: release-notes\n",[442,456,457],{"class":444,"line":315},[442,458,459],{"class":453},"description: Write release notes from merged pull requests in our house style. Use when asked to draft, write or prepare release notes or a changelog for a version.\n",[442,461,463],{"class":444,"line":462},4,[442,464,448],{"class":447},[442,466,468],{"class":444,"line":467},5,[442,469,470],{"emptyLinePlaceholder":341},"\n",[442,472,474],{"class":444,"line":473},6,[442,475,476],{"class":447},"# Release notes\n",[442,478,480],{"class":444,"line":479},7,[442,481,470],{"emptyLinePlaceholder":341},[442,483,485],{"class":444,"line":484},8,[442,486,487],{"class":447},"## When to use\n",[442,489,491],{"class":444,"line":490},9,[442,492,493],{"class":453},"The user asks for release notes, a changelog or \"what shipped\" for a version or date range.\n",[442,495,497],{"class":444,"line":496},10,[442,498,470],{"emptyLinePlaceholder":341},[442,500,502],{"class":444,"line":501},11,[442,503,504],{"class":447},"## Steps\n",[442,506,508,512,515,519,521,524],{"class":444,"line":507},12,[442,509,511],{"class":510},"s9osk","1.",[442,513,514],{"class":453}," List merged PRs for the range using the GitHub tools. Ignore PRs labelled ",[442,516,518],{"class":517},"sDLfK","`chore`",[442,520,255],{"class":453},[442,522,523],{"class":517},"`internal`",[442,525,526],{"class":453},".\n",[442,528,530,533],{"class":444,"line":529},13,[442,531,532],{"class":510},"2.",[442,534,535],{"class":453}," Group by: New, Improved, Fixed. One line each, present tense, no PR numbers in the line.\n",[442,537,539,542],{"class":444,"line":538},14,[442,540,541],{"class":510},"3.",[442,543,544],{"class":453}," Lead with the change that affects the most users.\n",[442,546,548,551,554,557],{"class":444,"line":547},15,[442,549,550],{"class":510},"4.",[442,552,553],{"class":453}," End with an \"Upgrade notes\" section only if any PR is labelled ",[442,555,556],{"class":517},"`breaking`",[442,558,526],{"class":453},[442,560,562],{"class":444,"line":561},16,[442,563,470],{"emptyLinePlaceholder":341},[442,565,567],{"class":444,"line":566},17,[442,568,569],{"class":447},"## Style\n",[442,571,573,576,579],{"class":444,"line":572},18,[442,574,575],{"class":453},"Short sentences. No exclamation marks. British spelling. Product names as in ",[442,577,578],{"class":517},"`docs\u002Fstyle.md`",[442,580,526],{"class":453},[11,582,583,584,587],{},"The ",[52,585,586],{},"description"," is doing real work. It is what the agent matches against, so it should say both what the skill does and when to use it, in the words a user would actually type.",[11,589,590],{},"A richer skill adds files:",[423,592,595],{"className":593,"code":594,"language":428},[426],"release-notes\u002F\n├── SKILL.md\n├── examples\u002F\n│   └── v2.4.md\n└── scripts\u002F\n    └── list_prs.sh\n",[52,596,594],{"__ignoreMap":313},[11,598,599,601,602,605,606,609],{},[52,600,54],{}," refers to them (\"see ",[52,603,604],{},"examples\u002Fv2.4.md"," for the expected format\", \"run ",[52,607,608],{},"scripts\u002Flist_prs.sh \u003Cfrom> \u003Cto>","\"). The agent reads or executes them only when it gets to that step.",[27,611,613],{"id":612},"installing-skills-in-claude-code","Installing skills in Claude Code",[11,615,616],{},"Claude Code looks for skills in three places.",[11,618,619,622,623,626],{},[18,620,621],{},"Project skills",", in ",[52,624,625],{},".claude\u002Fskills\u002F\u003Cname>\u002F"," inside the repository. Commit them, and everyone who clones the repo gets them. This is where team conventions belong.",[11,628,629,622,632,635],{},[18,630,631],{},"Personal skills",[52,633,634],{},"~\u002F.claude\u002Fskills\u002F\u003Cname>\u002F",". Yours across every project.",[11,637,638,641,642,645],{},[18,639,640],{},"Plugin skills",", delivered by plugins installed from a marketplace with ",[52,643,644],{},"\u002Fplugin",". Plugins bundle skills together with slash commands, hooks and MCP server definitions, which is how most third-party skills are distributed.",[11,647,648],{},"Once a skill is in place, there is nothing to enable. Type a request that matches its description and the agent uses it. Skills can also be invoked explicitly by name, and a skill can be marked so that only the user can trigger it, which is useful for anything with side effects.",[27,650,652],{"id":651},"skills-in-claudeai-and-the-api","Skills in Claude.ai and the API",[11,654,655],{},"In the Claude web and desktop apps, skills are managed under Settings, where you can enable Anthropic's built-in skills (documents, spreadsheets, presentations, PDFs) and upload your own as a zip of the skill folder. They run inside Claude's code execution sandbox.",[11,657,658],{},"Through the API, skills are attached to a request that uses the code execution container. You upload a skill once, reference it by id, and the model has it available for that request. This is how you give a production agent the same expertise a developer has in Claude Code.",[27,660,662],{"id":661},"writing-a-skill-that-works","Writing a skill that works",[11,664,665],{},"Most skills we see in the wild are too long, too vague, or both. The ones that reliably change the agent's behaviour share a few properties.",[11,667,668,671],{},[18,669,670],{},"The description is precise."," \"Helps with documents\" matches everything and nothing. \"Convert a DOCX contract into our redline format. Use when asked to redline, mark up or compare contract versions\" matches the right requests.",[11,673,674,677],{},[18,675,676],{},"The body is procedural."," Numbered steps, the order they should happen in, and what \"done\" looks like. The agent is good at following a procedure and bad at inferring one from a description of values.",[11,679,680,683],{},[18,681,682],{},"Decisions are explicit."," If the task has a fork (\"if the repo has a CHANGELOG.md, append to it; otherwise create one\"), write the fork. The agent will otherwise pick one at random and be confidently wrong half the time.",[11,685,686,689],{},[18,687,688],{},"Deterministic work goes in scripts."," Anything the model would do by generating code (parse a file format, call an API with the right pagination, compute a checksum) is more reliable as a script the skill runs. The model then orchestrates rather than reinvents.",[11,691,692,695,696,698],{},[18,693,694],{},"It is short enough to read."," Under 500 lines for ",[52,697,54],{},". Anything longer moves into reference files the agent opens on demand.",[11,700,701,704],{},[18,702,703],{},"It has an example."," One example of the expected output is worth a page of description.",[27,706,708],{"id":707},"where-to-find-skills","Where to find skills",[11,710,711],{},"Anthropic maintains a public repository of official skills on GitHub, covering document formats, presentations and a few developer workflows. Claude Code's plugin marketplaces are the main distribution channel for third-party skills. Community \"awesome\" lists collect more.",[11,713,714],{},"The rule is the same as for any code from the internet: read it before you install it. A skill can contain scripts, and scripts run. A skill's instructions can also tell the agent to do things you would not want (\"post the results to this webhook\"). For a team, that means an approved set rather than an open marketplace, which we come back to below.",[27,716,718],{"id":717},"skills-vs-the-alternatives","Skills vs the alternatives",[11,720,721,724,725,728,729,731],{},[18,722,723],{},"Skills vs CLAUDE.md."," A ",[52,726,727],{},"CLAUDE.md"," file is always loaded and should hold what the agent must always know: how to build, test, and the non-negotiable conventions. Skills hold task-specific knowledge that would bloat that file. If it applies to every session, put it in ",[52,730,727],{},"; if it applies to one kind of task, make it a skill.",[11,733,734,737],{},[18,735,736],{},"Skills vs slash commands."," Commands are user-triggered shortcuts for a prompt. Skills are triggered by the agent when relevant and can carry files and scripts. Many teams start with commands and graduate to skills when the prompt grows or needs supporting material.",[11,739,740,743],{},[18,741,742],{},"Skills vs subagents."," A subagent is a separate context with its own tools, used for isolation or parallelism. A skill is knowledge loaded into the current context. A subagent can use skills.",[11,745,746,749,750,220],{},[18,747,748],{},"Skills vs MCP."," MCP gives access to systems; skills give know-how. The full comparison is in ",[39,751,368],{"href":367},[27,753,755],{"id":754},"rolling-skills-out-to-a-team","Rolling skills out to a team",[11,757,758],{},"Three things change when skills go from one developer to a department.",[11,760,761,764],{},[18,762,763],{},"Provenance."," Which skills are approved, who wrote them, what scripts they contain. A curated internal marketplace, or a repository of reviewed skills, replaces \"install whatever the blog said\".",[11,766,767,770],{},[18,768,769],{},"Consistency."," The value of a skill for \"how we review PRs\" is that everyone's agent reviews PRs the same way. That only holds if the skill is the same everywhere and updated centrally.",[11,772,773,776],{},[18,774,775],{},"Policy."," A skill that says \"deploy to production after the tests pass\" should only run for people allowed to deploy. Skills do not enforce that; the tools they call do. This is where skills meet the gateway: the skill describes the procedure, the gateway decides whether this user's agent may call the deploy tool.",[11,778,779,780,220],{},"Walma AI Hub handles the last part. Approved skills and MCP servers are distributed through one signed client, policies decide who can call which tools, and every call is logged, inside the customer's own EU region. If you are standardising skills across a team, ",[39,781,782],{"href":309},"book a walkthrough",[784,785,786],"style",{},"html pre.shiki code .sIZOC, html code.shiki .sIZOC{--shiki-default:#79B8FF;--shiki-default-font-weight:bold}html pre.shiki code .s95oV, html code.shiki .s95oV{--shiki-default:#E1E4E8}html pre.shiki code .s9osk, html code.shiki .s9osk{--shiki-default:#FFAB70}html pre.shiki code .sDLfK, html code.shiki .sDLfK{--shiki-default:#79B8FF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}",{"title":313,"searchDepth":314,"depth":315,"links":788},[789,790,791,792,793,794,795,796,797],{"id":371,"depth":314,"text":372},{"id":384,"depth":314,"text":385},{"id":417,"depth":314,"text":418},{"id":612,"depth":314,"text":613},{"id":651,"depth":314,"text":652},{"id":661,"depth":314,"text":662},{"id":707,"depth":314,"text":708},{"id":717,"depth":314,"text":718},{"id":754,"depth":314,"text":755},"Agent skills are folders with a SKILL.md file that teach Claude how to do a specific task. Here is how they work in Claude Code, Claude.ai and the API, where to find good ones, how to build your own, and what to watch for before rolling them out to a team.",[800,803,806,809,812],{"q":801,"a":802},"What are Claude skills?","Skills are folders containing a SKILL.md file with instructions, plus optional scripts and reference files, that teach Claude how to perform a specific task. Claude loads a skill only when it is relevant, so you can have many without filling the context window.",{"q":804,"a":805},"How do I install a skill in Claude Code?","Put the skill folder in .claude\u002Fskills\u002F inside your project (shared with the team via git) or in ~\u002F.claude\u002Fskills\u002F for personal use. Skills can also come from plugins installed from a marketplace. Claude Code picks them up automatically.",{"q":807,"a":808},"Are skills the same as MCP?","No. Skills are instructions and scripts that tell the model how to do something. MCP servers give the model access to external tools and data. A skill often uses MCP tools to do its work. They are complementary.",{"q":810,"a":811},"Do skills work with ChatGPT, Codex or Cursor?","The SKILL.md format was published as an open specification called Agent Skills in late 2025, and Codex, Cursor, GitHub Copilot and Gemini CLI have adopted it. A well-written skill is portable across those tools, with some differences in how they are discovered.",{"q":813,"a":814},"Is there a Claude skills marketplace?","Anthropic publishes a set of official skills on GitHub, and Claude Code supports plugin marketplaces that bundle skills, commands and MCP servers. Several community directories exist. Treat skills like code: review what they run before installing.",{},0,"\u002Fguides\u002Fskills","11 min read",{"title":353,"description":798},"guides\u002Fskills\u002Findex","claude-skills-guide","LG3JmwILePIcGQB8BLmima6sXP2zfd1Sklu50acYdg0",{"id":4,"title":5,"author":6,"body":824,"date":324,"description":325,"extension":326,"faq":1019,"meta":1024,"navigation":341,"order":342,"path":343,"readTime":344,"seo":1025,"stem":346,"topic":347,"translationId":348,"updated":324,"__hash__":349},{"type":8,"value":825,"toc":1010},[826,828,834,836,842,850,852,930,932,936,938,940,942,944,946,950,968,970,972,980,982,990,992,994,996,998,1000,1004],[11,827,13],{},[11,829,16,830,21,832,25],{},[18,831,20],{},[18,833,24],{},[27,835,30],{"id":29},[11,837,33,838,37,840,43],{},[18,839,36],{},[39,841,42],{"href":41},[11,843,46,844,50,846,55,848,60],{},[18,845,49],{},[52,847,54],{},[39,849,59],{"href":58},[27,851,64],{"id":63},[66,853,854,864],{},[69,855,856],{},[72,857,858,860,862],{},[75,859],{},[75,861,36],{},[75,863,81],{},[83,865,866,874,882,890,898,906,914,922],{},[72,867,868,870,872],{},[88,869,90],{},[88,871,93],{},[88,873,96],{},[72,875,876,878,880],{},[88,877,101],{},[88,879,104],{},[88,881,107],{},[72,883,884,886,888],{},[88,885,112],{},[88,887,115],{},[88,889,118],{},[72,891,892,894,896],{},[88,893,123],{},[88,895,126],{},[88,897,129],{},[72,899,900,902,904],{},[88,901,134],{},[88,903,137],{},[88,905,140],{},[72,907,908,910,912],{},[88,909,145],{},[88,911,148],{},[88,913,151],{},[72,915,916,918,920],{},[88,917,156],{},[88,919,159],{},[88,921,162],{},[72,923,924,926,928],{},[88,925,167],{},[88,927,170],{},[88,929,173],{},[27,931,177],{"id":176},[11,933,180,934],{},[18,935,183],{},[11,937,186],{},[11,939,189],{},[11,941,192],{},[27,943,196],{"id":195},[11,945,199],{},[11,947,202,948,206],{},[52,949,205],{},[208,951,952,954,958,962,964],{},[211,953,213],{},[211,955,216,956,220],{},[52,957,219],{},[211,959,223,960,227],{},[52,961,226],{},[211,963,230],{},[211,965,233,966,237],{},[52,967,236],{},[11,969,240],{},[27,971,244],{"id":243},[11,973,247,974,251,976,255,978,259],{},[52,975,250],{},[52,977,254],{},[52,979,258],{},[11,981,262],{},[264,983,984,986,988],{},[211,985,268],{},[211,987,271],{},[211,989,274],{},[11,991,277],{},[27,993,281],{"id":280},[11,995,284],{},[27,997,288],{"id":287},[11,999,291],{},[11,1001,294,1002,298],{},[52,1003,297],{},[11,1005,301,1006,306,1008,311],{},[39,1007,305],{"href":304},[39,1009,310],{"href":309},{"title":313,"searchDepth":314,"depth":315,"links":1011},[1012,1013,1014,1015,1016,1017,1018],{"id":29,"depth":314,"text":30},{"id":63,"depth":314,"text":64},{"id":176,"depth":314,"text":177},{"id":195,"depth":314,"text":196},{"id":243,"depth":314,"text":244},{"id":280,"depth":314,"text":281},{"id":287,"depth":314,"text":288},[1020,1021,1022,1023],{"q":329,"a":330},{"q":332,"a":333},{"q":335,"a":336},{"q":338,"a":339},{},{"title":5,"description":325},[1027],{"loc":1028,"href":367},"en",1789134650275]