> 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/operator-agent/skills/creating-skills.md). # Creating Skills Skills are created by owners and builders. You can build one yourself from the Skills page, or ask the agent to create one during a chat session and approve its proposal. ### Creating a Skill from the Skills Page 1. **Open the Skills page** in your project and click **New**. 2. **Give the skill a name** that describes the task (e.g. "Order Processing", "Quote Generation"). 3. **Pick an icon**: choose an emoji or upload an image so the skill is easy to spot. 4. **Write the instructions**: the core of the skill. Tell the agent what to do, how to behave, and what output to produce. 5. **Add a description** of what the skill does. If you leave it empty, one is generated for you when you leave the page. Your changes save automatically as a draft while you work. ### Creating a Skill from Agent Chat You can also ask the agent to create or change a skill during a conversation. For example: * *"Create a skill that processes incoming purchase orders from email attachments"* * *"Turn what you just did into a reusable skill"* The agent never saves anything itself. Instead it proposes the change and asks for your approval: 1. **A review card appears in the chat** summarising what the agent wants to do. 2. **Open the full review in the side panel** to see exactly what would change, including a preview of the version that would be created or published. 3. **Approve or cancel**. Depending on the proposal, the approve button reads **Create draft**, **Create Skill**, **Save changes**, **Publish changes**, **Discard draft**, or **Revert**. Choose **Cancel** and nothing changes. > ⚠️ The agent can propose creating, updating, publishing, discarding, reverting, or deleting a skill, but none of it happens until you approve it in the review. ### Adding Components Once a skill exists, add components that give it access to systems and data. Open the **Add Component** dialog on the skill page and pick a tab: * **Workflows**: let the skill run an existing workflow from your project. You can require approval before the workflow runs. * **Files**: upload reference documents such as templates or examples. You can view and edit a file's content from the skill page. * **Connections**: connect apps the skill needs (e.g. Gmail, Business Central, Slack). Decide which account the skill uses when operators run it: a **shared account** that everyone uses, or a **personal account** where each operator connects their own before using the skill. * **Data Tables**: structured data the agent can read from or write to. You can require approval before the skill writes to a table. * **Knowledge Bases**: document collections the skill can search. Useful for skills that need reference data such as product catalogues, pricing rules, policy documents, or technical manuals. ### Drafts, Publishing, and Versions A skill can have a draft alongside its published version, so you can work on changes without affecting what operators use: * **Drafts**: editing a skill creates or updates its draft while the published version keeps running unchanged. Skills with unpublished changes show a **Draft** badge. * **Testing**: owners and builders can **Test** a draft in an agent session before publishing it, even when a published version already exists. * **Publishing**: publish the draft to make the changes live. A skill needs a description and instructions before it can be published, and each publish creates a new version. * **Discarding**: throw the draft away and keep the published version as it is. Discarded changes cannot be recovered. * **Reverting**: go back to an earlier published version. The selected version replaces the current draft, and you publish it to make it live. Skills keep a version history, so you can see what changed and when, and revert to an earlier version if needed. > 💡 The agent only ever runs published, enabled skills. Draft changes are never used in operators' sessions until you publish them. ### Editing a Skill During a Chat Session You can adjust a skill without leaving a conversation: * **Open the skill's panel**: click the skill in the chat to open its detail panel beside the conversation. Owners and builders can edit the instructions and components there directly. * **Ask the agent**: describe the change you want, then approve the agent's proposal from the review card. ### Managing Skills From a skill's page, you can: * **Edit** the skill's name, icon, description, instructions, and components * **Enable / disable** the skill for the current project * **View version history** and revert to an earlier version * **Delete** the skill if it's no longer needed --- # 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/operator-agent/skills/creating-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.