> For the complete documentation index, see [llms.txt](https://docs.lleverage.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.lleverage.ai/run-operations/agent/agent-context/skills.md). # Skills Understand Skills as reusable operating methods the Agent can use, create and improve around recurring operational work. Skills are reusable operating methods that teach the Agent how your organisation handles a recurring kind of work. When you give the Agent a task, it can identify relevant Skills and use them as part of the work. You can also choose a Skill explicitly when you know which capability you want to use. A task may use more than one Skill when several methods are needed. For example, a purchasing Skill might tell the Agent which supplier information to check, which system is authoritative, what exceptions matter, what action to take and when to involve a person. ## Using Skills in Agent work The most important place to experience a Skill is the Agent session itself. A Skill gives the Agent a reusable method without turning the task into a rigid Workflow. During a session, the Agent can select an appropriate Skill from the Project. You can also invoke one deliberately when you want a particular method followed. Learn how to use Skills ## What a Skill can use A Skill can define the operating method and use the Project capabilities needed to carry it out, including: * **instructions and checks**: how to approach the task and what a good result looks like; * **Connections**: authorised access to systems such as email, ERP or collaboration tools; * **Workflows**: defined execution the Agent can use for stable parts of the task; * **Knowledge**: maintained reference material the Agent can search; * **Tables**: structured project data the Agent can query or update when appropriate; * **Files**: templates, examples and working material. A Skill does **not** own or grant credentials. Connections provide system authority; the Skill tells the Agent how and when to use that authority. ## The Agent can build its own Skills Most Skill building is intended to happen from the Agent itself, in the context of the work you are trying to teach it. A builder can give the Agent a representative task, examples, rules and corrections. The Agent writes or updates the Skill, then presents the change through the dedicated Skill-building review UI before it becomes the maintained project capability. That makes the practical loop: **do representative work → explain the expected method → Agent writes or updates the Skill → review → test on representative cases → correct → retest** ### Test Skill changes before publishing Eligible Project builders can use the **Tests** surface from Skill detail and Skill proposal views to run representative evaluations while reviewing a Skill. Access follows the normal Project and acting-user permission boundary. Treat test results as evidence for the review. A passing test does not silently publish a draft, and failing or negative evidence should lead to an explicit correction/review decision rather than an automatic production rewrite. The separate Intelligence → Skills surface remains the canonical inventory and direct-edit surface for maintained Skills. It is not the only, or usually the primary, authoring journey. Repeated operator corrections are evidence that a Skill may need improvement. They do not mean a production Skill silently rewrites itself without review. What you can create or change depends on your project access. See Roles and Permissions for the current access model. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.lleverage.ai/run-operations/agent/agent-context/skills.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.