> ## Documentation Index > Fetch the complete documentation index at: https://dataleap-469bb19d-docs-from-commits-2026-08-19-upload-limits.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Skills > A skill is a reusable part of an agent: define a standard once — your ICP, your pricing methodology, a report format — and every agent that needs it follows it. Diagram: three agents side by side — Lead Qualifier, Outreach Writer, Pipeline Report — each with the same ICP Definition skill as a building block in its workflow Diagram: three agents side by side — Lead Qualifier, Outreach Writer, Pipeline Report — each with the same ICP Definition skill as a building block in its workflow ## What makes a good skill | Good skill | Example | | ----------------------------- | ---------------------------------------------------------------------------------------------------------- | | **Encode knowledge** | Your Ideal Customer Profile — every agent qualifies leads the same way. | | **Standardize a methodology** | Your internal pricing methodology — defined once, applied identically by every agent that prices anything. | | **Define your data** | What each metric or field means, and what it's called in your systems — shared analytics language. | | **Standardize outputs** | How a report should be structured and worded. | | **Lock the design** | The look for documents and dashboards — always on-brand. | The signal that something should be a skill: you catch yourself typing the same definition or procedure into a second agent — "wait, I'm redoing this work." Inside, a skill is just clear instructions — here's a real one: An open skill showing the ICP Scoring Methodology with scoring criteria tables and an Edit dropdown offering Edit in Chat and Edit Manually ## Create a skill Two ways — out of an agent you already built, or from scratch: You built an agent and notice one part of it is reusable — just tell Dataleap: *"Create a skill out of the scoring logic for new candidates."* It drafts the skill from what you built, proposes name, description, and content, and saves once you confirm. A chat where the user asks to create a skill out of the scoring logic and Dataleap drafts the Candidate Scoring Model skill Click **Add Skill** and pick how: **Create in Chat**, **Create Manually**, or **Import Markdown** (if the standard already exists as a document). The Add Skill dropdown with the options Create in Chat, Create Manually, and Import Markdown There's no hard size limit on a skill — but every rule in it costs tokens on every run that loads it. Keep skills as lean as possible while still doing the job. ## Use a skill Agents don't guess which skills to use — you tell them. When building or refining an agent, reference the skill explicitly: > "When it comes to pricing, use the *Pricing Methodology* skill." From then on, that step of the workflow runs on the skill's definition. And when the methodology changes, you update the skill once — every agent using it follows. Be proactive: Dataleap won't suggest skills on its own. Spotting the reusable part is your job; packaging and reusing it is one sentence. ## Share a skill Skills have three levels of reach: * **Personal** — your own skills, visible only to you. * **Team** — share a skill into a [team space](/share/team-spaces) (it has its own **Skills** tab); every teammate's agents can use it. * **Org-wide** — publish it for the whole company. The classic case: company-wide metric definitions, so every agent anyone builds speaks the same data language. The Skills section listing skills grouped into Personal, Team, and Org, with an Add Skill button and workspace filters See how skills work in the product. **Skills vs. knowledge:** knowledge files give one agent *context* (documents, facts — see [Add knowledge](/build/add-knowledge)); a skill teaches *procedure* and can be reused by many agents. Next: cut costs on deterministic workflows. Where team skills live.